diff --git "a/data/results_20260528_150751_cc695181.jsonl" "b/data/results_20260528_150751_cc695181.jsonl"
deleted file mode 100644--- "a/data/results_20260528_150751_cc695181.jsonl"
+++ /dev/null
@@ -1,52 +0,0 @@
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 2048, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 120.9, "vram_cliff_tokens": null, "avg_power_w": 49.6, "max_power_w": 54.9, "avg_gpu_temp_c": 71.1, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 80.0, "max_cpu_temp_c": 84.6, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 225.51, "p50_ttft_ms": 211.02, "p99_ttft_ms": 343.16, "avg_itl_ms": 30.57, "p50_itl_ms": 29.08, "p99_itl_ms": 75.88, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:42:03.175733+00:00", "submitted_at": "2026-05-28T15:07:51.319037+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "e2bc1623a8ff43cb93d88f7f3534558f", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "f8c57b503ae80647059bc9ff3c9799f49383a9f3ce68c932a64d5dd2a519fbfe", "result_fingerprint": "1e3e64d83700024f67316eda090c3bfb9a10c391971cc7b2e4f8bf506fa17547", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 2048, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 86.84, "vram_cliff_tokens": null, "avg_power_w": 47.2, "max_power_w": 48.6, "avg_gpu_temp_c": 73.1, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 84.5, "max_cpu_temp_c": 85.1, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 142.66, "p50_ttft_ms": 135.44, "p99_ttft_ms": 186.96, "avg_itl_ms": 22.49, "p50_itl_ms": 21.61, "p99_itl_ms": 43.24, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:42:59.792926+00:00", "submitted_at": "2026-05-28T15:07:51.320674+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "29e1281643d24dc590bcde7e4b845f89", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "9b8721693fee4be6862f4a572740ccb21a9e2db0b4fe2d403b3c9da31e7599ab", "result_fingerprint": "d54660bb09824d510fb14096d8412a8bc113f413b7b272188ec1c20a0a3d9952", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 2048, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 55.68, "vram_cliff_tokens": null, "avg_power_w": 47.5, "max_power_w": 48.8, "avg_gpu_temp_c": 73.7, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 85.0, "max_cpu_temp_c": 85.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 88.07, "p50_ttft_ms": 89.91, "p99_ttft_ms": 116.98, "avg_itl_ms": 17.66, "p50_itl_ms": 17.56, "p99_itl_ms": 21.68, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:43:43.779485+00:00", "submitted_at": "2026-05-28T15:07:51.322275+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "feac737cd42b454f9a96e399fe81741c", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "54097f547ee9851dfc6ea297945656c43407509ad8f83d14e6a67fad8bde6364", "result_fingerprint": "5211a48b30db7886b540200f550385b119a298622a47fd01296f6cf5b86c4d0c", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 2048, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 124.66, "vram_cliff_tokens": null, "avg_power_w": 52.9, "max_power_w": 57.7, "avg_gpu_temp_c": 75.8, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 85.8, "max_cpu_temp_c": 87.0, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 192.65, "p50_ttft_ms": 177.91, "p99_ttft_ms": 262.9, "avg_itl_ms": 29.5, "p50_itl_ms": 28.38, "p99_itl_ms": 49.28, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:44:59.559237+00:00", "submitted_at": "2026-05-28T15:07:51.323910+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "5e1d027c3d9b430a850949e30dc6a373", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "eb2309c6dcae66862d6c3ceb5ce09cfae7a5a3ffcc287b32cae20f42b9259efd", "result_fingerprint": "c5aaa7db370308a2cb556662a17fdb1c012f7ac883ae74508d9330317146eeb9", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 2048, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 81.56, "vram_cliff_tokens": null, "avg_power_w": 47.4, "max_power_w": 49.1, "avg_gpu_temp_c": 74.8, "max_gpu_temp_c": 76.0, "avg_cpu_temp_c": 86.6, "max_cpu_temp_c": 86.9, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 140.51, "p50_ttft_ms": 133.04, "p99_ttft_ms": 192.78, "avg_itl_ms": 23.87, "p50_itl_ms": 21.63, "p99_itl_ms": 43.88, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:45:59.613198+00:00", "submitted_at": "2026-05-28T15:07:51.325513+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "55d866a7bbf94bd8b071ac545a17e06b", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "ca1c178433006cafa32adfc46cd01fa594989274e4441c262bdd6957a0584032", "result_fingerprint": "74f6e970944a90b27f4443e70915cc6bda9bda5b58c32e7df69d2fe00db9c181", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 2048, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 55.72, "vram_cliff_tokens": null, "avg_power_w": 48.1, "max_power_w": 49.2, "avg_gpu_temp_c": 75.0, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 87.2, "max_cpu_temp_c": 88.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 86.91, "p50_ttft_ms": 87.88, "p99_ttft_ms": 118.95, "avg_itl_ms": 17.65, "p50_itl_ms": 17.53, "p99_itl_ms": 21.4, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:46:43.550224+00:00", "submitted_at": "2026-05-28T15:07:51.327132+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "a80995dd842f4280b19b000b754592a1", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "c49bb522f3105f162a17ab29e0970bd9535b5e10b9f30b5c2112b8362c0d7f77", "result_fingerprint": "7425b5221119f2b6b324e53108a23c5e1d8b3867c78489fac4e06334388819d9", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 8192, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 194.42, "vram_cliff_tokens": null, "avg_power_w": 42.9, "max_power_w": 57.4, "avg_gpu_temp_c": 71.6, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 80.4, "max_cpu_temp_c": 85.2, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 420.29, "p50_ttft_ms": 416.21, "p99_ttft_ms": 706.01, "avg_itl_ms": 76.52, "p50_itl_ms": 67.84, "p99_itl_ms": 207.08, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:50:18.391979+00:00", "submitted_at": "2026-05-28T15:07:51.328781+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "cfd13abe23934c3ea0e364f767118ce4", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "8955468924f444e2cfb43bf1a76f375f3887e0f750a1f6f9b559b3f39690aab7", "result_fingerprint": "d381959393e7ff730ad29de9a231ec268dabe2ced2683bf2c159527c0eb2e547", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 8192, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 95.12, "vram_cliff_tokens": null, "avg_power_w": 48.0, "max_power_w": 63.7, "avg_gpu_temp_c": 73.4, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 83.5, "max_cpu_temp_c": 86.4, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 355.9, "p50_ttft_ms": 376.42, "p99_ttft_ms": 530.01, "avg_itl_ms": 81.32, "p50_itl_ms": 82.68, "p99_itl_ms": 146.9, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:53:39.885208+00:00", "submitted_at": "2026-05-28T15:07:51.330421+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "5754b1666a364b56b9e6010f89f9d31e", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "207129851f7035c7968f356f46371460eb59e9df65e467e09e5a6dd95760bb2d", "result_fingerprint": "99f7652865a4a563d1c444b14cad8bb0544932d894e2a62d84785159ffc07ae3", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 8192, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 70.65, "vram_cliff_tokens": null, "avg_power_w": 45.2, "max_power_w": 54.5, "avg_gpu_temp_c": 72.4, "max_gpu_temp_c": 75.0, "avg_cpu_temp_c": 82.7, "max_cpu_temp_c": 83.8, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 246.71, "p50_ttft_ms": 232.86, "p99_ttft_ms": 419.97, "avg_itl_ms": 55.1, "p50_itl_ms": 66.27, "p99_itl_ms": 90.75, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:55:55.234920+00:00", "submitted_at": "2026-05-28T15:07:51.332021+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "d2d471168e6147e096c212e58f1856fd", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "4e7a3a0a85c0dc9959389b56a73975e4bd43c56d7fabe7de1f3cf0ffe8234e89", "result_fingerprint": "d0159f368ff7b8816156d53782c5916b80efc2487775ffa93519c88324191a06", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 8192, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 58.62, "vram_cliff_tokens": null, "avg_power_w": 44.6, "max_power_w": 48.6, "avg_gpu_temp_c": 72.2, "max_gpu_temp_c": 75.0, "avg_cpu_temp_c": 82.8, "max_cpu_temp_c": 83.6, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 171.63, "p50_ttft_ms": 144.8, "p99_ttft_ms": 275.93, "avg_itl_ms": 33.38, "p50_itl_ms": 41.1, "p99_itl_ms": 44.73, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:57:18.900754+00:00", "submitted_at": "2026-05-28T15:07:51.333590+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "3ebd646365df4af5acf90646d7c5978f", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "aeaf12b5e57ba1e2a24c6e93ade6fc6873a847d01feff133026e4d74c3ed9a69", "result_fingerprint": "bcd8c0fc7032ef2cae4b3015fd81f7e74c59035b2ab47b9ad278327b870be735", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 8192, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 55.71, "vram_cliff_tokens": null, "avg_power_w": 47.5, "max_power_w": 48.9, "avg_gpu_temp_c": 73.5, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 83.4, "max_cpu_temp_c": 84.1, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 88.53, "p50_ttft_ms": 91.05, "p99_ttft_ms": 125.46, "avg_itl_ms": 17.62, "p50_itl_ms": 17.46, "p99_itl_ms": 21.42, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T12:58:00.818371+00:00", "submitted_at": "2026-05-28T15:07:51.335150+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "8135f12b1203405091f0ceaa73d34fb0", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "3136f7140ba5f65717e4322aa509469ab2e336c201778563d896fa949943b1c1", "result_fingerprint": "2e1ac29190e124c2eefe7f54c6489505dd40a90c611ea40420569c8695e487e4", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 8192, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 207.19, "vram_cliff_tokens": null, "avg_power_w": 45.2, "max_power_w": 57.2, "avg_gpu_temp_c": 72.2, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 82.1, "max_cpu_temp_c": 84.9, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 468.36, "p50_ttft_ms": 411.68, "p99_ttft_ms": 857.42, "avg_itl_ms": 73.03, "p50_itl_ms": 65.59, "p99_itl_ms": 217.44, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:01:07.658546+00:00", "submitted_at": "2026-05-28T15:07:51.336763+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "c95a0a30a3e34df0aa38e59c3003408c", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "4b5ee79ef86f0579c33ada577744d68542101c9b9c99511a049ab09cb5c818fe", "result_fingerprint": "06b1e2801745cca914a4a60fea9b1aed59f5ee9b9944532289521eeb474fbcd4", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 8192, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 100.1, "vram_cliff_tokens": null, "avg_power_w": 49.0, "max_power_w": 63.0, "avg_gpu_temp_c": 73.2, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 83.3, "max_cpu_temp_c": 84.9, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 355.45, "p50_ttft_ms": 348.07, "p99_ttft_ms": 564.51, "avg_itl_ms": 77.94, "p50_itl_ms": 82.71, "p99_itl_ms": 145.78, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:04:20.915277+00:00", "submitted_at": "2026-05-28T15:07:51.338404+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "6d5e0abb4fee4a27887922050aed6678", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "dfd2402728efc807077adc3103cbd4e33582ab01785f007a32f2443282360832", "result_fingerprint": "7da7c9810244af31166a0441878e92b083275a15bc6cc4c85e898d4fdca46d6e", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 8192, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 93.33, "vram_cliff_tokens": null, "avg_power_w": 48.3, "max_power_w": 56.4, "avg_gpu_temp_c": 73.6, "max_gpu_temp_c": 76.0, "avg_cpu_temp_c": 83.3, "max_cpu_temp_c": 85.5, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 204.37, "p50_ttft_ms": 198.78, "p99_ttft_ms": 315.2, "avg_itl_ms": 40.66, "p50_itl_ms": 30.23, "p99_itl_ms": 86.08, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:06:02.505953+00:00", "submitted_at": "2026-05-28T15:07:51.339994+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "7e61249db80e4ccc8ed2eecdfa3d9e0b", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "175cacac4a10b963eecd62fc32abafcadc6a5ed2db2ec5fcccafdd26d4f23e29", "result_fingerprint": "92cda4b0ab8d526a0ff5b41af67972cb229189c88979f62336eea7fe36e67c03", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 8192, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 65.91, "vram_cliff_tokens": null, "avg_power_w": 45.3, "max_power_w": 48.8, "avg_gpu_temp_c": 72.4, "max_gpu_temp_c": 73.0, "avg_cpu_temp_c": 82.3, "max_cpu_temp_c": 82.8, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 162.37, "p50_ttft_ms": 148.18, "p99_ttft_ms": 231.21, "avg_itl_ms": 29.49, "p50_itl_ms": 22.0, "p99_itl_ms": 44.95, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:07:16.842409+00:00", "submitted_at": "2026-05-28T15:07:51.341551+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "585b592bd9c34f1bac4276826a2ea450", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "99b4549caada85f82a3ef037fb6b2d76c36be4cb926e9600e4251e96e34eb87a", "result_fingerprint": "53b53ca48a59ae0d791d6e46ccb5103c40d0ba8b79c1e8d4d3faf4f4ab16ee8f", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 8192, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 55.84, "vram_cliff_tokens": null, "avg_power_w": 47.5, "max_power_w": 48.1, "avg_gpu_temp_c": 72.9, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 82.6, "max_cpu_temp_c": 83.1, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 86.26, "p50_ttft_ms": 90.19, "p99_ttft_ms": 124.69, "avg_itl_ms": 17.63, "p50_itl_ms": 17.48, "p99_itl_ms": 21.61, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:08:01.955827+00:00", "submitted_at": "2026-05-28T15:07:51.343115+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "87aec346a3ba405e89f904340b41da1e", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "87c5a0a740c5a74108898cc68fd99a94891995fbe54b045df6b1b52ba1822c2c", "result_fingerprint": "be0e1caec60bcc0781936a37a028fe68ce6fa6e619114eb14ee4475297e64905", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 258.46, "vram_cliff_tokens": null, "avg_power_w": 45.7, "max_power_w": 59.8, "avg_gpu_temp_c": 71.9, "max_gpu_temp_c": 78.0, "avg_cpu_temp_c": 77.8, "max_cpu_temp_c": 84.3, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 580.76, "p50_ttft_ms": 580.08, "p99_ttft_ms": 1286.36, "avg_itl_ms": 116.49, "p50_itl_ms": 96.56, "p99_itl_ms": 364.33, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:13:29.623739+00:00", "submitted_at": "2026-05-28T15:07:51.344825+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "28bc923b3fd74334a36a89b40f2c39be", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "9126887d3dddc582a8ae344c95a3e0bc3b391993bc5c5453322ebdb8d87b7d1f", "result_fingerprint": "f93ab2acc0cc2f19b4a069d63974bedf567939c92db796905902e61c63837679", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 106.79, "vram_cliff_tokens": null, "avg_power_w": 46.0, "max_power_w": 61.5, "avg_gpu_temp_c": 72.1, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 82.4, "max_cpu_temp_c": 85.6, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 634.0, "p50_ttft_ms": 587.62, "p99_ttft_ms": 1034.45, "avg_itl_ms": 145.52, "p50_itl_ms": 147.76, "p99_itl_ms": 273.56, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:19:38.545980+00:00", "submitted_at": "2026-05-28T15:07:51.346460+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "94848cb0070d4c02b59451ee3d341583", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "519cfa40045d8259aa4fa682be8433a36f658981c04fe2f077f18463aa010680", "result_fingerprint": "5a8fa7705c0627508bedced4b8c293654cc8f60e3002ad206f4ab9f5d4879559", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 89.92, "vram_cliff_tokens": null, "avg_power_w": 47.4, "max_power_w": 65.7, "avg_gpu_temp_c": 72.7, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 83.4, "max_cpu_temp_c": 85.9, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 380.67, "p50_ttft_ms": 364.79, "p99_ttft_ms": 643.61, "avg_itl_ms": 86.02, "p50_itl_ms": 85.1, "p99_itl_ms": 146.03, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:23:16.859979+00:00", "submitted_at": "2026-05-28T15:07:51.348057+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "285bfedf543e42389fc3bbe290a2bfb1", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "7405a029b9f0fa5d6a9fad0571215967bd42cd7310719b452f671d5216608047", "result_fingerprint": "3919b1549150a036e70ab0434b3076aebcdd019744209a76214ed5a9779a226c", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 68.93, "vram_cliff_tokens": null, "avg_power_w": 44.6, "max_power_w": 55.1, "avg_gpu_temp_c": 71.8, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 82.0, "max_cpu_temp_c": 84.1, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 245.77, "p50_ttft_ms": 245.1, "p99_ttft_ms": 321.95, "avg_itl_ms": 56.11, "p50_itl_ms": 51.69, "p99_itl_ms": 92.67, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:25:36.703862+00:00", "submitted_at": "2026-05-28T15:07:51.349656+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "867b502d48ec4c2083af785eb2fb74a7", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "4f7410a23a14730931241c55eedee34379a81480eaaba8e47a3adb456d44d576", "result_fingerprint": "ebb64597539b3e6c7ef458230714a258b41b690c4288a648cbfca9bbfa82ce08", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 68.17, "vram_cliff_tokens": null, "avg_power_w": 45.5, "max_power_w": 49.6, "avg_gpu_temp_c": 71.8, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 81.8, "max_cpu_temp_c": 82.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 134.73, "p50_ttft_ms": 136.7, "p99_ttft_ms": 179.72, "avg_itl_ms": 28.8, "p50_itl_ms": 21.93, "p99_itl_ms": 44.89, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:26:51.362456+00:00", "submitted_at": "2026-05-28T15:07:51.351240+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "1b0d72f6448d49c0a253a7b8c4fef0ca", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "aaf470ef86e177daa2bfb1a3626b5c7f0b68f090971050e7273e2562f9458f04", "result_fingerprint": "582e0046feb45a09a0d4b1ba01e3b17856d4786e703ebc285319895bfdefee92", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 55.76, "vram_cliff_tokens": null, "avg_power_w": 47.4, "max_power_w": 48.2, "avg_gpu_temp_c": 72.5, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 82.4, "max_cpu_temp_c": 82.8, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 86.86, "p50_ttft_ms": 95.36, "p99_ttft_ms": 126.34, "avg_itl_ms": 17.66, "p50_itl_ms": 17.51, "p99_itl_ms": 21.64, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:27:37.679436+00:00", "submitted_at": "2026-05-28T15:07:51.352804+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "c21fcdf73c6746398db0e3a57002f05a", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "b009d8827ecf61af76073567766cac6abf286ed659da138f7fbe86f19e09db02", "result_fingerprint": "4001054640fa1b2779c63f866118eefc7b4b8f41ff25c33ed66e3ab38e98746b", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 254.64, "vram_cliff_tokens": null, "avg_power_w": 47.6, "max_power_w": 60.0, "avg_gpu_temp_c": 72.5, "max_gpu_temp_c": 78.0, "avg_cpu_temp_c": 79.0, "max_cpu_temp_c": 84.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 625.9, "p50_ttft_ms": 517.36, "p99_ttft_ms": 1190.98, "avg_itl_ms": 116.1, "p50_itl_ms": 95.7, "p99_itl_ms": 332.91, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:32:41.873973+00:00", "submitted_at": "2026-05-28T15:07:51.354499+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "9c04126c65554e799041e4924ee1ef8d", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "a1a8233dbcbbfdc8679ef5af07e74038e8b78f8366c7a08709bc552b44ea19ff", "result_fingerprint": "8cce6c0744cb1f05ba6da3e0c6d7dd2695174cc45f01bf4b325b36ea85dc6c6b", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 116.21, "vram_cliff_tokens": null, "avg_power_w": 47.5, "max_power_w": 57.1, "avg_gpu_temp_c": 72.6, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 82.2, "max_cpu_temp_c": 85.8, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 559.54, "p50_ttft_ms": 603.42, "p99_ttft_ms": 811.86, "avg_itl_ms": 132.35, "p50_itl_ms": 142.51, "p99_itl_ms": 277.42, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:38:15.860786+00:00", "submitted_at": "2026-05-28T15:07:51.356142+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "8f4b8a50af5348bf807a9645f6ba812e", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "c01879ec707835e5db76bc35b9b43cc83bfecce5f6aff05ea24cb476d473d97b", "result_fingerprint": "b8dc4078b031065b757d9458f39e498a76f9518e6e4b8be45396b269027ad9d6", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 79.77, "vram_cliff_tokens": null, "avg_power_w": 46.5, "max_power_w": 65.8, "avg_gpu_temp_c": 72.0, "max_gpu_temp_c": 76.0, "avg_cpu_temp_c": 82.1, "max_cpu_temp_c": 85.5, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 408.44, "p50_ttft_ms": 405.24, "p99_ttft_ms": 787.09, "avg_itl_ms": 95.8, "p50_itl_ms": 99.92, "p99_itl_ms": 158.86, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:42:18.372091+00:00", "submitted_at": "2026-05-28T15:07:51.357753+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "3e77070fd9a64255914cb00ef0e1bbbc", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "d7e00c5b6e0fe3b2316088ea7cf9f6af078f68775a8c8775f81f5c34a0998b3b", "result_fingerprint": "6ece47c08250ec527bdf149aef7a0b251368917f6b20c87812f27ac3b0175c3a", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 68.35, "vram_cliff_tokens": null, "avg_power_w": 45.1, "max_power_w": 55.2, "avg_gpu_temp_c": 71.5, "max_gpu_temp_c": 73.0, "avg_cpu_temp_c": 81.7, "max_cpu_temp_c": 82.8, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 263.72, "p50_ttft_ms": 267.44, "p99_ttft_ms": 391.09, "avg_itl_ms": 57.24, "p50_itl_ms": 65.94, "p99_itl_ms": 87.74, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:44:45.884650+00:00", "submitted_at": "2026-05-28T15:07:51.359335+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "46c06464993d44d0b6be693e9752ae14", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "02c4354a7ed78d24c27d898ce47537c923407aa92d22430e5c2a639c298f9d06", "result_fingerprint": "2373a97205988bb14ef270fc95997863aa8e8cfb17307238ab884483c6aaaf32", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 58.0, "vram_cliff_tokens": null, "avg_power_w": 44.5, "max_power_w": 48.6, "avg_gpu_temp_c": 71.2, "max_gpu_temp_c": 72.0, "avg_cpu_temp_c": 81.5, "max_cpu_temp_c": 82.1, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 155.19, "p50_ttft_ms": 149.75, "p99_ttft_ms": 207.98, "avg_itl_ms": 33.91, "p50_itl_ms": 41.25, "p99_itl_ms": 44.99, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:46:13.152094+00:00", "submitted_at": "2026-05-28T15:07:51.360898+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "fe40055d75004350b725c870722700a2", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "fb54e723e0f672e17f4daf529e7ee325729d402e4716986ef90e41e617223f53", "result_fingerprint": "c5d98591c7a8043897574665ec3d5abba4262c489936cad59f015e05410f27fb", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 55.65, "vram_cliff_tokens": null, "avg_power_w": 47.2, "max_power_w": 48.2, "avg_gpu_temp_c": 72.3, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 82.4, "max_cpu_temp_c": 83.0, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 90.27, "p50_ttft_ms": 99.01, "p99_ttft_ms": 128.57, "avg_itl_ms": 17.67, "p50_itl_ms": 17.54, "p99_itl_ms": 21.8, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:46:58.625302+00:00", "submitted_at": "2026-05-28T15:07:51.362481+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "88bf993b5b894063bd3cfb740355a206", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "4643a0a630d8a2161a0f837861153ed79c031e0f19b33ff49313c9d4ef7c8292", "result_fingerprint": "9625667df4bce065241193a1ee9be0146bb35641222649806fde968f3433c1c8", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 261.27, "vram_cliff_tokens": null, "avg_power_w": 45.0, "max_power_w": 67.5, "avg_gpu_temp_c": 71.2, "max_gpu_temp_c": 78.0, "avg_cpu_temp_c": 77.5, "max_cpu_temp_c": 83.8, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 607.13, "p50_ttft_ms": 551.49, "p99_ttft_ms": 1392.02, "avg_itl_ms": 114.86, "p50_itl_ms": 96.12, "p99_itl_ms": 344.41, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:52:24.706721+00:00", "submitted_at": "2026-05-28T15:07:51.364151+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "7e26e2073c614862bb00e7658e73c9c4", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "3e9e004271499a69a34e1261398824d030d88f7d685cba8efd2f50ea98da2985", "result_fingerprint": "015e3fbbe1001951f939aba3240de200a52d8800308c4c0738a45639bc1cad78", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 102.17, "vram_cliff_tokens": null, "avg_power_w": 45.7, "max_power_w": 56.9, "avg_gpu_temp_c": 71.6, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 81.6, "max_cpu_temp_c": 84.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 658.53, "p50_ttft_ms": 619.66, "p99_ttft_ms": 1093.83, "avg_itl_ms": 150.93, "p50_itl_ms": 169.94, "p99_itl_ms": 287.55, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T13:58:46.901431+00:00", "submitted_at": "2026-05-28T15:07:51.365778+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "74ca66ff749a4c6c88a4318f96b616e2", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "7ae6d3249fac7fafb0d3da24fe73ab927974dc7481702a657990cd3cd9caf607", "result_fingerprint": "dddc6b108607b19c2c7d8cfe3ed501940bd1fe83d1eb23a47ca9ed6754a0c90b", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 84.35, "vram_cliff_tokens": null, "avg_power_w": 46.6, "max_power_w": 64.6, "avg_gpu_temp_c": 72.0, "max_gpu_temp_c": 76.0, "avg_cpu_temp_c": 83.2, "max_cpu_temp_c": 85.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 387.55, "p50_ttft_ms": 383.65, "p99_ttft_ms": 751.07, "avg_itl_ms": 92.84, "p50_itl_ms": 99.56, "p99_itl_ms": 146.84, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:02:43.262686+00:00", "submitted_at": "2026-05-28T15:07:51.367387+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "a58d782c94ca4858919b66b2bf4c9917", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "9b8590c6f4e5a874ed8b6aa4184039919f72c00d0e9a15501b0f8a33f85d363c", "result_fingerprint": "11c440d750bd90aabaef067afad2dad16eb295843098d277cc3ad85bacf847ad", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 72.26, "vram_cliff_tokens": null, "avg_power_w": 45.8, "max_power_w": 56.3, "avg_gpu_temp_c": 72.4, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 84.9, "max_cpu_temp_c": 85.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 275.66, "p50_ttft_ms": 235.13, "p99_ttft_ms": 501.85, "avg_itl_ms": 54.03, "p50_itl_ms": 65.51, "p99_itl_ms": 88.25, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:04:58.726785+00:00", "submitted_at": "2026-05-28T15:07:51.368961+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "367858cda3c341a89535049466a9321f", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "19b9bad4c92b2312df6775e5f8bdadd1426300a844f2973670e3df21df53185b", "result_fingerprint": "655ba7ac0f3377c4f2381706200bfdb88da503fb961c6831ad38d4c07d65d350", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 57.67, "vram_cliff_tokens": null, "avg_power_w": 44.7, "max_power_w": 48.2, "avg_gpu_temp_c": 72.3, "max_gpu_temp_c": 73.0, "avg_cpu_temp_c": 85.1, "max_cpu_temp_c": 85.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 141.3, "p50_ttft_ms": 139.64, "p99_ttft_ms": 211.07, "avg_itl_ms": 32.92, "p50_itl_ms": 41.14, "p99_itl_ms": 44.84, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:06:20.567993+00:00", "submitted_at": "2026-05-28T15:07:51.370552+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "c356f0a48b5f4767960f9e1c67f60188", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "02eaf068c0d6480a846c11ab2c2025b6a18b7490c45e2a5344ab94b784dd402d", "result_fingerprint": "5350997c462b9aa17394f8d783a51ed8b195b917019e10b6d8eeb0eabdd2dc4f", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 55.77, "vram_cliff_tokens": null, "avg_power_w": 47.5, "max_power_w": 48.5, "avg_gpu_temp_c": 73.1, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 85.4, "max_cpu_temp_c": 85.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 85.4, "p50_ttft_ms": 92.18, "p99_ttft_ms": 121.49, "avg_itl_ms": 17.65, "p50_itl_ms": 17.52, "p99_itl_ms": 21.44, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:07:06.314914+00:00", "submitted_at": "2026-05-28T15:07:51.372126+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "7cea0234ac6a4ee3a64118eeee294a4f", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "b1af0d699f02184e5f1b6a6834d646b116376a44c4f2d43c7d13a98d11333e29", "result_fingerprint": "391e63f14645458a5bb3e5afaa0eb7113bc7d9debddcba675a9661866000da08", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 256.93, "vram_cliff_tokens": null, "avg_power_w": 47.6, "max_power_w": 60.4, "avg_gpu_temp_c": 72.8, "max_gpu_temp_c": 78.0, "avg_cpu_temp_c": 79.0, "max_cpu_temp_c": 84.1, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 629.23, "p50_ttft_ms": 527.65, "p99_ttft_ms": 1202.96, "avg_itl_ms": 116.83, "p50_itl_ms": 97.69, "p99_itl_ms": 347.41, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:12:12.008843+00:00", "submitted_at": "2026-05-28T15:07:51.373785+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "499a67ca809c42e99c1bf3d533291a0f", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "d65b27661fc9697dd585857d9feaa552c0f79e1358f16f46e54405cb54396933", "result_fingerprint": "1d2323e3b5bc2b7779335373f1ec164128506a08571b90662467608a98a8a6d6", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 102.51, "vram_cliff_tokens": null, "avg_power_w": 47.2, "max_power_w": 61.0, "avg_gpu_temp_c": 72.9, "max_gpu_temp_c": 78.0, "avg_cpu_temp_c": 83.1, "max_cpu_temp_c": 86.9, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 684.87, "p50_ttft_ms": 601.99, "p99_ttft_ms": 1293.19, "avg_itl_ms": 147.65, "p50_itl_ms": 154.17, "p99_itl_ms": 247.66, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:18:26.358949+00:00", "submitted_at": "2026-05-28T15:07:51.375119+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "ae867e81713a4d0ca92d5535ff40fffd", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "584a4b9a72cdda57d900f1176170763e6a80afa9ef1fea4f3e9aca758ed4ccc2", "result_fingerprint": "1417e88bc8935024210eb713e9ac48b304d1c858f6436df8072aeb75166b10a0", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 74.2, "vram_cliff_tokens": null, "avg_power_w": 45.8, "max_power_w": 64.3, "avg_gpu_temp_c": 72.1, "max_gpu_temp_c": 76.0, "avg_cpu_temp_c": 82.2, "max_cpu_temp_c": 84.5, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 441.25, "p50_ttft_ms": 427.82, "p99_ttft_ms": 874.57, "avg_itl_ms": 104.54, "p50_itl_ms": 116.34, "p99_itl_ms": 161.8, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:22:51.111797+00:00", "submitted_at": "2026-05-28T15:07:51.376150+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "14d1728ec885483188841ddd2fed8854", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "aff4ca12ffe9f1edad5b3eaffa0d08d0e9c3697587103a47e518dd2e0bd238d6", "result_fingerprint": "3f76ca1f7ca09870a29364e0bd167228eafb787e190459c26f4b048ab3cda46a", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 65.97, "vram_cliff_tokens": null, "avg_power_w": 44.8, "max_power_w": 55.9, "avg_gpu_temp_c": 71.6, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 82.3, "max_cpu_temp_c": 84.6, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 272.35, "p50_ttft_ms": 253.07, "p99_ttft_ms": 436.69, "avg_itl_ms": 59.09, "p50_itl_ms": 66.63, "p99_itl_ms": 91.44, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:25:19.908809+00:00", "submitted_at": "2026-05-28T15:07:51.377132+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "213ab1a60ebf4cca802b624b8cfa0715", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "09bb3d197cf06a65ddd10ad64c24583ccd6a837a9c0836b2d7589e3394d67543", "result_fingerprint": "945e4fbb4bf1e8c5d1de9035e6df8645e3ded1209fac8530f713fe26802111dc", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 60.44, "vram_cliff_tokens": null, "avg_power_w": 45.1, "max_power_w": 48.5, "avg_gpu_temp_c": 72.4, "max_gpu_temp_c": 73.0, "avg_cpu_temp_c": 85.0, "max_cpu_temp_c": 85.6, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 136.27, "p50_ttft_ms": 138.33, "p99_ttft_ms": 187.83, "avg_itl_ms": 31.53, "p50_itl_ms": 40.29, "p99_itl_ms": 44.75, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:26:40.021150+00:00", "submitted_at": "2026-05-28T15:07:51.378095+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "157f99184b034e948769647da3a9d228", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "9371f0e243e52f6697e725614bd4b7c355d2e03f76d1aad63dcea9b37dccdff9", "result_fingerprint": "55b35754d42d95fe3fa1147ff628de6e8152dd94830b13d59ae18fe5a61f0c6e", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 65536, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 55.78, "vram_cliff_tokens": null, "avg_power_w": 47.5, "max_power_w": 48.2, "avg_gpu_temp_c": 73.4, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 85.4, "max_cpu_temp_c": 85.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 84.35, "p50_ttft_ms": 79.5, "p99_ttft_ms": 125.93, "avg_itl_ms": 17.66, "p50_itl_ms": 17.53, "p99_itl_ms": 21.67, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:27:26.306838+00:00", "submitted_at": "2026-05-28T15:07:51.379043+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "42c5f0468370458a90de1c7f8c462c95", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "f7cc70e4fc618daed346cb61cc96657b934a4efb58081c3870220ab477161196", "result_fingerprint": "f53050912bf445575669a2dc700a73d6631ccc5846c6c729612ddd7280a2b30c", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 256.13, "vram_cliff_tokens": null, "avg_power_w": 45.4, "max_power_w": 63.2, "avg_gpu_temp_c": 71.9, "max_gpu_temp_c": 78.0, "avg_cpu_temp_c": 78.1, "max_cpu_temp_c": 84.5, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 562.76, "p50_ttft_ms": 503.35, "p99_ttft_ms": 1173.03, "avg_itl_ms": 116.6, "p50_itl_ms": 95.65, "p99_itl_ms": 337.71, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:32:57.369849+00:00", "submitted_at": "2026-05-28T15:07:51.380046+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "653d15c8d9934ed396d99fd9baa22cb2", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "5c7a6729ba15f5c2ed9ffb9904c1d62d9cf2ac7a44bd4b1c9d463f508c65d0c0", "result_fingerprint": "23dd73912ba818016e32f7faf14c07ac99ed94ffad8994f5cc0df11f45727fa9", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 97.97, "vram_cliff_tokens": null, "avg_power_w": 46.7, "max_power_w": 55.2, "avg_gpu_temp_c": 72.5, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 82.5, "max_cpu_temp_c": 85.6, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 639.03, "p50_ttft_ms": 561.62, "p99_ttft_ms": 1160.71, "avg_itl_ms": 155.93, "p50_itl_ms": 150.51, "p99_itl_ms": 286.85, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:39:28.594856+00:00", "submitted_at": "2026-05-28T15:07:51.381027+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "45e615016178437db1cacf6904eff1e6", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "850c668d985a65455081bd5afefecc591a5690cb90a276ae06fbfccc341991c0", "result_fingerprint": "c9a98baf52474ca19a33133fc38ded57e608bbb9a0301ebc079098a5f1003341", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 82.95, "vram_cliff_tokens": null, "avg_power_w": 46.6, "max_power_w": 64.9, "avg_gpu_temp_c": 72.5, "max_gpu_temp_c": 76.0, "avg_cpu_temp_c": 82.3, "max_cpu_temp_c": 84.6, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 417.32, "p50_ttft_ms": 377.98, "p99_ttft_ms": 799.67, "avg_itl_ms": 92.54, "p50_itl_ms": 98.7, "p99_itl_ms": 156.18, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:43:23.550089+00:00", "submitted_at": "2026-05-28T15:07:51.381982+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "a56d5b94bbe24618b0dc5de238731ff1", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "ddc7f3a9d8d0697deefc141a43d8f5839762783a29bbbd50b8b9a34b00aa0182", "result_fingerprint": "c492c1bdc7e3c8ae0a66e7931b21f4a29e2f31a3978dfb1432aab73de841c48e", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 68.88, "vram_cliff_tokens": null, "avg_power_w": 45.1, "max_power_w": 55.2, "avg_gpu_temp_c": 71.8, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 81.9, "max_cpu_temp_c": 83.1, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 245.94, "p50_ttft_ms": 255.08, "p99_ttft_ms": 356.12, "avg_itl_ms": 55.33, "p50_itl_ms": 50.14, "p99_itl_ms": 92.76, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:45:42.014218+00:00", "submitted_at": "2026-05-28T15:07:51.382923+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "ad22779a44e34159abb893c58759dd70", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "6d0c4ea34b84ae3493797c883fa81926e0dfd6e468cdc9f965445de18a36d88f", "result_fingerprint": "333e04eaeb6a81b9d542d345ecf32b95c2bbf3e6606f601b0e9419fc3c626aad", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 68.42, "vram_cliff_tokens": null, "avg_power_w": 45.5, "max_power_w": 48.8, "avg_gpu_temp_c": 72.1, "max_gpu_temp_c": 73.0, "avg_cpu_temp_c": 82.3, "max_cpu_temp_c": 83.1, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 139.18, "p50_ttft_ms": 145.04, "p99_ttft_ms": 185.2, "avg_itl_ms": 28.68, "p50_itl_ms": 22.03, "p99_itl_ms": 44.86, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:46:55.777895+00:00", "submitted_at": "2026-05-28T15:07:51.383871+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "c63ecde06b7041fda5433206c7c9add9", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "3bdec6daaa80de279f95cf62e6faaa0d7d0f7d55f2868e2a4b97282c30734a5a", "result_fingerprint": "07c5d2882532ec3df2abe1063936a0c476e910f849797c8e2087a01ed381188f", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 55.82, "vram_cliff_tokens": null, "avg_power_w": 47.4, "max_power_w": 48.3, "avg_gpu_temp_c": 72.7, "max_gpu_temp_c": 73.0, "avg_cpu_temp_c": 82.9, "max_cpu_temp_c": 83.8, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 86.29, "p50_ttft_ms": 85.43, "p99_ttft_ms": 126.52, "avg_itl_ms": 17.62, "p50_itl_ms": 17.46, "p99_itl_ms": 21.65, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:47:40.151400+00:00", "submitted_at": "2026-05-28T15:07:51.384814+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "91062378ecf344d1be6fb4188b6cac52", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "f8e26a322ca9640ca6a89f810b790f6c62220a42282e2e373c36740dcfecf346", "result_fingerprint": "55926310c8109763d8ec3ec59fbf00c06db64a9f13649b1c233ca8405fcd0c4b", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 249.16, "vram_cliff_tokens": null, "avg_power_w": 47.5, "max_power_w": 72.6, "avg_gpu_temp_c": 72.9, "max_gpu_temp_c": 79.0, "avg_cpu_temp_c": 79.1, "max_cpu_temp_c": 84.6, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 933.75, "p50_ttft_ms": 844.38, "p99_ttft_ms": 2634.17, "avg_itl_ms": 116.07, "p50_itl_ms": 97.86, "p99_itl_ms": 324.39, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:52:49.680771+00:00", "submitted_at": "2026-05-28T15:07:51.385807+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "a5281e30ae3f42708e0984e37911a93a", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "a0f9bdef12e8855f01c98464b23784340982db7e93a6888bf16d06c57976fafa", "result_fingerprint": "d885363fcf4b3a8f8832a56db8a32f3ef7d611826b7f04f11923f21792746ecb", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 99.36, "vram_cliff_tokens": null, "avg_power_w": 46.4, "max_power_w": 54.4, "avg_gpu_temp_c": 72.4, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 82.6, "max_cpu_temp_c": 85.5, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 603.41, "p50_ttft_ms": 580.84, "p99_ttft_ms": 941.88, "avg_itl_ms": 155.41, "p50_itl_ms": 165.17, "p99_itl_ms": 272.86, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T14:59:22.457102+00:00", "submitted_at": "2026-05-28T15:07:51.386785+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "b4f84c1aa6204b0392d3ec600d1c69c8", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "bd4215d058bce4542f6287f978b9406a2bb01b9cdcca4a71f7383d72e945d05a", "result_fingerprint": "419df4b0a191a6cf4bf5d45fb3a4dad90665e99cc2d07fe6bdf0a8bf41d89fe9", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 88.24, "vram_cliff_tokens": null, "avg_power_w": 47.3, "max_power_w": 65.6, "avg_gpu_temp_c": 72.7, "max_gpu_temp_c": 77.0, "avg_cpu_temp_c": 85.0, "max_cpu_temp_c": 87.6, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 376.3, "p50_ttft_ms": 334.47, "p99_ttft_ms": 769.87, "avg_itl_ms": 87.39, "p50_itl_ms": 83.41, "p99_itl_ms": 152.31, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T15:03:05.025104+00:00", "submitted_at": "2026-05-28T15:07:51.387734+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "ea358e93d531469fbb931d21dd5133a2", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "552fa74ff033aa185a3233f3275dee3e0822cf368817f1e9ea411402b65502aa", "result_fingerprint": "04f750b67a048cfe3cf22cd7ef637be5339db9b38633acf8c830cbc65382a4c4", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 73.6, "vram_cliff_tokens": null, "avg_power_w": 45.6, "max_power_w": 54.1, "avg_gpu_temp_c": 72.4, "max_gpu_temp_c": 74.0, "avg_cpu_temp_c": 84.3, "max_cpu_temp_c": 85.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 261.09, "p50_ttft_ms": 228.44, "p99_ttft_ms": 451.79, "avg_itl_ms": 52.61, "p50_itl_ms": 49.85, "p99_itl_ms": 87.31, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T15:05:14.832027+00:00", "submitted_at": "2026-05-28T15:07:51.388680+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "28bf7c79df1441d5b2bfdd9cad79d392", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "423c2f96c2f8c81d83892e7a3a536dceaffeb1fe15e3989d2ce16dfb92ecff38", "result_fingerprint": "2c6fcfb9715cfdc7d75278349accdaadb784cffaaa3efcf5a41688fad183156d", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 53.18, "vram_cliff_tokens": null, "avg_power_w": 44.1, "max_power_w": 48.2, "avg_gpu_temp_c": 71.9, "max_gpu_temp_c": 73.0, "avg_cpu_temp_c": 84.0, "max_cpu_temp_c": 84.6, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 146.13, "p50_ttft_ms": 146.42, "p99_ttft_ms": 222.46, "avg_itl_ms": 36.39, "p50_itl_ms": 41.38, "p99_itl_ms": 45.06, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T15:06:46.199731+00:00", "submitted_at": "2026-05-28T15:07:51.389625+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "43490ad77547499ea31d060ffa724c51", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "857e8b5fddf0fa83764605d5e9ad619197becfc79cb3b1aa83960044c581c829", "result_fingerprint": "819d294cbe0dc4d80c66286cd816c81739544811c0f5ac197310a12fa54ea88a", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}
-{"run_type": "all", "model": "unsloth/gemma-4-26B-A4B-it-GGUF/gemma-4-26B-A4B-it-UD-IQ4_XS.gguf", "model_base": "gemma-4-26B-A4B-it-UD", "quant": "IQ4_XS", "model_org": "unsloth", "model_repo": "unsloth/gemma-4-26B-A4B-it-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "NVIDIA GB10", "gpu_vram_gb": 119.6, "gpu_driver": "580.159.03", "gpu_count": 1, "gpu_names": "NVIDIA GB10", "gpu_total_vram_gb": 119.6, "unified_memory": true, "gpu_compute_capability": "12.1", "gpu_pcie_gen": 1, "gpu_pcie_width": 16, "gpu_power_limit_w": null, "backends": "CUDA", "cpu_model": null, "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "llm_flags": "{}", "llm_flags_label": null, "extra_flags_raw": null, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 55.92, "vram_cliff_tokens": null, "avg_power_w": 47.4, "max_power_w": 48.6, "avg_gpu_temp_c": 72.8, "max_gpu_temp_c": 73.0, "avg_cpu_temp_c": 84.9, "max_cpu_temp_c": 85.7, "avg_fan_speed_rpm": 2, "max_fan_speed_rpm": 2, "avg_ttft_ms": 85.06, "p50_ttft_ms": 95.44, "p99_ttft_ms": 127.81, "avg_itl_ms": 17.61, "p50_itl_ms": 17.46, "p99_itl_ms": 21.6, "quality_score": null, "context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null, "qualitative": {"context_rot_score": null, "context_rot_accuracy_by_length": null, "context_rot_accuracy_by_depth": null, "context_rot_accuracy_by_needle": null, "multi_needle_score": null, "multi_needle_accuracy_by_length": null, "tool_selection_accuracy": null, "parameter_accuracy": null, "parameter_hallucination_rate": null, "parse_success_rate": null, "no_call_accuracy": null, "overall_tool_accuracy": null, "knowledge_accuracy_mean": null, "knowledge_accuracy_std": null, "answer_relevancy_mean": null, "coherence_mean": null, "quality_composite_score": null, "memory_accuracy": null, "mt_bench_score": null, "cases_evaluated": null, "cases_skipped_context": null}, "quantitative": null, "meta": null, "os_system": "Linux", "os_release": "6.17.0-1018-nvidia", "os_machine": "aarch64", "os_distro": "Ubuntu", "os_distro_version": "24.04", "cpu_cores": "20", "ram_total_gb": 119.6, "submitter": "Paul Lee", "timestamp": "2026-05-28T15:07:31.652492+00:00", "submitted_at": "2026-05-28T15:07:51.390566+00:00", "schema_version": "0.10.0", "benchmark_version": "0.1.0", "suite_run_id": "3fbb5f4037624aab899f4b3904089e87", "submission_id": "3fbb5f4037624aab899f4b3904089e87", "row_id": "29f5a621a53f420cabd8d2e7a78140c8", "machine_fingerprint": "45bd15b6fc7710689087295e975db7ab49775c3d1e6688222884df41351ab4b3", "run_fingerprint": "7e3807a13c6e43edc967d10ef9c3894e1fb505c0b30562a3bf813b61557dfe2a", "result_fingerprint": "adc37dce43130b09922d1800a984488d6bcfed461401fabf56634b34a0633e4e", "source_file_sha256": "87fba7981d2579ea165e1275c271a3a89a1a95c8b14cc6b7e7b203b3305da288", "tags": null}