Hugging Face's logo

Datasets:
paulplee
/
ppb-results

Tasks:
Tabular Regression
Tabular Classification
Modalities:
Tabular
Text
Formats:
parquet
Languages:
English
Size:
10K - 100K
Tags:
benchmarking
llama-cpp
llm-inference
local-llm
gpu
quantization
Libraries:
Datasets
Dask
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
paulplee commited on
Commit
378d39a
·
verified ·
1 Parent(s): 323d0ce

Upload data/results_20260425_093103_7c852dae.jsonl with huggingface_hub

Browse files
Files changed (1) hide show
  1. data/results_20260425_093103_7c852dae.jsonl +46 -0
data/results_20260425_093103_7c852dae.jsonl ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 8192, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 70.35, "avg_power_w": 33.6, "max_power_w": 38.8, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1103.33, "p50_ttft_ms": 855.71, "p99_ttft_ms": 4045.35, "avg_itl_ms": 215.16, "p50_itl_ms": 206.94, "p99_itl_ms": 485.95, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T04:29:39.677113+00:00", "submitted_at": "2026-04-25T09:31:02.607170+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "962a825cb74146fbaa2f89cba64177c5", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "6cc086e51545d5882eb86ec26c438cff7c7527ce90aec8cc54e6438810a9d6b8", "result_fingerprint": "1eb3a79cc6d2447b328845e753bc8ac3e616b16e59245d07dfdf2e11c5fb3b7e", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
2
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 8192, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 50.36, "avg_power_w": 31.6, "max_power_w": 38.2, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 700.87, "p50_ttft_ms": 593.11, "p99_ttft_ms": 1520.3, "avg_itl_ms": 152.23, "p50_itl_ms": 150.74, "p99_itl_ms": 242.86, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T04:36:02.655845+00:00", "submitted_at": "2026-04-25T09:31:02.611051+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "9508b36491eb4cb49e933e82c80b3264", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "d22282afda4ba5a49399ea0672260c36d9d4a4317db167bddb3dd95301b82a4b", "result_fingerprint": "a6841d6a5f1786da4d7d2dfe4e8fbcf800dfd6d29fa310219f810deb2e46f01c", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
3
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 8192, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 46.21, "avg_power_w": 32.7, "max_power_w": 38.2, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 511.38, "p50_ttft_ms": 346.5, "p99_ttft_ms": 947.15, "avg_itl_ms": 83.22, "p50_itl_ms": 85.05, "p99_itl_ms": 107.7, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T04:39:37.204443+00:00", "submitted_at": "2026-04-25T09:31:02.614187+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "b405248764aa42828ac99178ec58320a", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "e8242f974fc8fce50d6117d0827d854a0d70b61304ce3fa24b1922e7978ca8f5", "result_fingerprint": "840bd5fcef6b7be451caa5e81adab868f5d861126c1071409897c9f1815025b9", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
4
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 8192, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 45.07, "avg_power_w": 33.3, "max_power_w": 38.8, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 218.31, "p50_ttft_ms": 224.75, "p99_ttft_ms": 380.74, "avg_itl_ms": 43.59, "p50_itl_ms": 48.99, "p99_itl_ms": 55.14, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T04:41:28.697183+00:00", "submitted_at": "2026-04-25T09:31:02.617031+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "76fa00188ae946648bda3f218da89f28", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "289d4e32679c83b751ae2125c52ec06eac02ef4db25f648f4cdfffd0e0ea6c77", "result_fingerprint": "c779e66accc6525e3df9828807b19ab67aa06e145f50b17e7c57a7953a432925", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
5
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 8192, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 40.07, "avg_power_w": 34.2, "max_power_w": 38.9, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 142.55, "p50_ttft_ms": 161.25, "p99_ttft_ms": 316.09, "avg_itl_ms": 24.45, "p50_itl_ms": 24.43, "p99_itl_ms": 26.8, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T04:42:33.966609+00:00", "submitted_at": "2026-04-25T09:31:02.619643+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "3cb4f3e928914f76879a065b70d47d37", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "a1c66a284c0de442ccebce87f6fb0e2fc0207433d28ff1857c30ed7230f7bc6b", "result_fingerprint": "fc9b9f1d821420de7993f0efaf43542f80afb926615e3813020dbf554e0eda1a", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
6
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 8192, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 67.16, "avg_power_w": 31.3, "max_power_w": 37.7, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 990.35, "p50_ttft_ms": 832.98, "p99_ttft_ms": 2930.7, "avg_itl_ms": 227.54, "p50_itl_ms": 220.07, "p99_itl_ms": 442.57, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T04:51:59.765608+00:00", "submitted_at": "2026-04-25T09:31:02.622192+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "861d501bba4b466bb3986860c19075e1", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "5775e5e4ae6fdc83d8274fd6e27bccfa26c0801f199e9b951efdd4bd1fb74208", "result_fingerprint": "12548ed7ac0fd7711e56efb301d63139fe5214eaad383a585c8d197658eb3c3f", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
7
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 8192, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 50.59, "avg_power_w": 37.0, "max_power_w": 39.6, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 653.07, "p50_ttft_ms": 593.4, "p99_ttft_ms": 1275.49, "avg_itl_ms": 151.73, "p50_itl_ms": 147.06, "p99_itl_ms": 227.83, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T04:58:22.899168+00:00", "submitted_at": "2026-04-25T09:31:02.624633+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "8a0237c65d2e44279169bde8f217d717", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "49e9c0280e55832a6f8b92c3e7e4620a39d8e456f4af463ea4b114bbeeb24fbf", "result_fingerprint": "30451247bfe7384028cfe2e704de2b3e29c253d74c4c3014f4450e73c93dfdf7", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
8
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 8192, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 44.05, "avg_power_w": 36.0, "max_power_w": 39.3, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 407.11, "p50_ttft_ms": 305.84, "p99_ttft_ms": 951.85, "avg_itl_ms": 86.33, "p50_itl_ms": 96.79, "p99_itl_ms": 115.67, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T05:01:52.914065+00:00", "submitted_at": "2026-04-25T09:31:02.627521+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "ee4b38b147ee4dad907cc8d8175f16f1", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "6a4824827b83775c8734bb71bfa397952e3bbc3535b19799e8c1e9f2d7087b15", "result_fingerprint": "2051c8b3ff3aba7ee54f52ea1ccd4ad1d9dde8559b4cf8f9fef1754f20d535ab", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
9
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 8192, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 41.04, "avg_power_w": 34.8, "max_power_w": 37.7, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 225.93, "p50_ttft_ms": 223.49, "p99_ttft_ms": 480.71, "avg_itl_ms": 47.49, "p50_itl_ms": 49.92, "p99_itl_ms": 51.6, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T05:03:58.322316+00:00", "submitted_at": "2026-04-25T09:31:02.630627+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "264e11ae218c47ea8145dddf0c8d5d76", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "3495cb02204bad1bd271166206cbfdfb5d014d1be1c2f8fc06625d74a5be850d", "result_fingerprint": "d9cb283bb3d259ec361b8b0df996cf09b1c5632b017412c68e65bffc35655c7a", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
10
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 8192, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 41.13, "avg_power_w": 35.8, "max_power_w": 37.5, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 131.76, "p50_ttft_ms": 149.53, "p99_ttft_ms": 278.66, "avg_itl_ms": 23.87, "p50_itl_ms": 24.04, "p99_itl_ms": 25.18, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T05:05:04.789517+00:00", "submitted_at": "2026-04-25T09:31:02.633452+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "aabeb8207e4c4c9986d76986a9d300ad", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "bd4e09636337665d38a9b3846771736de7c4f7ac2d4e0e9670477f19c751d8f4", "result_fingerprint": "ad9240d2bd89ec33729c47123ea9ec724e96d465d50923782a852783f304524a", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
11
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 75.14, "avg_power_w": 33.6, "max_power_w": 41.9, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1770.58, "p50_ttft_ms": 1350.66, "p99_ttft_ms": 5583.7, "avg_itl_ms": 407.45, "p50_itl_ms": 396.57, "p99_itl_ms": 955.1, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T05:22:35.929451+00:00", "submitted_at": "2026-04-25T09:31:02.636031+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "ef129e6f4b614e18ac417d5f7d641506", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "bd92a7f0942fe8a177d9b6323d3882fc26b53797d3801acd8a6177e77a1abd9e", "result_fingerprint": "bfb6a479dd0b7308a0fa0e20d845e716fd12d646187d0655dfc96b14eddad201", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
12
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 53.0, "avg_power_w": 36.0, "max_power_w": 39.8, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1430.32, "p50_ttft_ms": 1231.55, "p99_ttft_ms": 3774.87, "avg_itl_ms": 286.68, "p50_itl_ms": 284.38, "p99_itl_ms": 515.13, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T05:34:38.494802+00:00", "submitted_at": "2026-04-25T09:31:02.638520+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "d72584ed12094f1aad42eb84bab33ff4", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "c75ea651f718f8de0a035a31862912c1746f1681e91debe8a82da14d9cbf0d93", "result_fingerprint": "5c716aa05d3dc58ca953e63d11c6a0d6b721eb1086607cebd0cb8fb30cc2a89a", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
13
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 45.95, "avg_power_w": 35.2, "max_power_w": 40.1, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 723.81, "p50_ttft_ms": 584.89, "p99_ttft_ms": 1667.42, "avg_itl_ms": 168.89, "p50_itl_ms": 166.59, "p99_itl_ms": 240.1, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T05:41:43.069979+00:00", "submitted_at": "2026-04-25T09:31:02.640949+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "e37f367aae8b41cfac0055b5e9727a95", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "00019837007795a33fa8fe5b79ad30dd18ac558d4d37a8665b82d5037d3580f7", "result_fingerprint": "0ddcb988afc29e963222cddf8f2b470b833d433cd806373311d7aa8ee84a24c5", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
14
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 43.38, "avg_power_w": 35.4, "max_power_w": 37.1, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 398.02, "p50_ttft_ms": 335.46, "p99_ttft_ms": 857.8, "avg_itl_ms": 89.31, "p50_itl_ms": 85.53, "p99_itl_ms": 118.41, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T05:45:29.669818+00:00", "submitted_at": "2026-04-25T09:31:02.643211+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "fea83545b637402b821efc4c9f3e5a6e", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "102477616fa1de5499eaadb8455152f0aa185ae2fdf26ef8dc83eb663dc89fa0", "result_fingerprint": "45ebc124036801e28cb8ffd5724f2b7f4738da33c1fb34d219510c3793effaf4", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
15
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 41.03, "avg_power_w": 35.0, "max_power_w": 37.5, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 228.62, "p50_ttft_ms": 223.49, "p99_ttft_ms": 488.21, "avg_itl_ms": 47.8, "p50_itl_ms": 49.87, "p99_itl_ms": 52.15, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T05:47:31.841392+00:00", "submitted_at": "2026-04-25T09:31:02.645345+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "ec9dfc35e1b74e75acf7083a1b598af5", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "28db38042212d9a608a6eb084f4b846100057b0f6b91eb534e977e1ac51fa1cd", "result_fingerprint": "a239ade7761f9992d09cee54b1329a95e22428ac8e42c5fc141a543b38b932fe", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
16
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 41.09, "avg_power_w": 35.8, "max_power_w": 37.2, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 134.39, "p50_ttft_ms": 156.38, "p99_ttft_ms": 266.18, "avg_itl_ms": 23.86, "p50_itl_ms": 24.02, "p99_itl_ms": 25.67, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T05:48:35.511637+00:00", "submitted_at": "2026-04-25T09:31:02.647368+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "7245c1bc29174e13b8a7fd8fffda6895", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "5708966278125867736da508e282695007fd612cc20497f1914a07932c06fe26", "result_fingerprint": "10a893785c38795077bc5276a9aa4ec64b1d5060bd99febb7c5f80b39474604d", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
17
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 75.6, "avg_power_w": 34.1, "max_power_w": 40.9, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 2124.5, "p50_ttft_ms": 1802.8, "p99_ttft_ms": 9159.2, "avg_itl_ms": 405.05, "p50_itl_ms": 393.31, "p99_itl_ms": 952.74, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T06:05:46.910448+00:00", "submitted_at": "2026-04-25T09:31:02.649370+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "e96edd40bc6e4d2d9d5d2fbff6d71bfb", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "dba55d3169c9db2435036d1e1525f9c9e9309f1ccd0994872a548dd000039a3e", "result_fingerprint": "b364ba9cf8679f756ba36ff4d8796fc0f844f798b75851891dd52c241ecaa4d0", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
18
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 50.27, "avg_power_w": 36.0, "max_power_w": 40.4, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1385.71, "p50_ttft_ms": 1146.37, "p99_ttft_ms": 2927.38, "avg_itl_ms": 305.57, "p50_itl_ms": 302.2, "p99_itl_ms": 544.91, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T06:18:32.231274+00:00", "submitted_at": "2026-04-25T09:31:02.651359+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "6f7e426de91f4809816b557443da3d8f", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "ffc1eb250e06b7c768a64aec6bb59def8ae3590f09a9c951ecd3962512a7ef1f", "result_fingerprint": "f30db0ef2933f584ba151f5a0b2e6678bcce8fbfffe284a6d9ebb5c24bb21deb", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
19
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 44.58, "avg_power_w": 35.5, "max_power_w": 39.5, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 811.96, "p50_ttft_ms": 609.88, "p99_ttft_ms": 1697.23, "avg_itl_ms": 173.46, "p50_itl_ms": 173.62, "p99_itl_ms": 220.5, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T06:25:53.677737+00:00", "submitted_at": "2026-04-25T09:31:02.653307+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "845d372149df4f799fb0f70a6e617490", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "98a568212542dd570d10855efa4bc1c1c832b6ea5912357b623c32c4bfc50ef1", "result_fingerprint": "41b9cdbadfa803e9901781d9a9b8ac854a19899c3a67ac8ee8f8053bc6feaaa5", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
20
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 42.6, "avg_power_w": 35.2, "max_power_w": 37.1, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 407.46, "p50_ttft_ms": 364.92, "p99_ttft_ms": 936.7, "avg_itl_ms": 91.61, "p50_itl_ms": 95.75, "p99_itl_ms": 117.85, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T06:29:46.838431+00:00", "submitted_at": "2026-04-25T09:31:02.655224+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "11167b003ac94b649877e5e7b300aa18", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "6896ec12d2283425526d67620dfcbee476308a1841a121ad4c570189536992cf", "result_fingerprint": "fd0bbb4d4c6389bc6838f6d9d2339ef7f2831d7c4af14d2580130a6bdc63448d", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
21
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 40.91, "avg_power_w": 34.9, "max_power_w": 37.1, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 226.58, "p50_ttft_ms": 217.37, "p99_ttft_ms": 483.26, "avg_itl_ms": 47.55, "p50_itl_ms": 49.66, "p99_itl_ms": 52.8, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T06:31:52.086390+00:00", "submitted_at": "2026-04-25T09:31:02.657000+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "c3bda37b3ede4beda5192bb4f365ee76", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "d5555f0773e36b5b32760a3df7377c6b2e20ec2e9e9ca5cab84d0d27a2673ae0", "result_fingerprint": "f24d62d37ae1688fee2dd1a090f99aada81808279b9638ec8add546662014b3c", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
22
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 16384, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 42.03, "avg_power_w": 36.8, "max_power_w": 39.4, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 136.06, "p50_ttft_ms": 150.94, "p99_ttft_ms": 277.9, "avg_itl_ms": 23.34, "p50_itl_ms": 23.26, "p99_itl_ms": 25.39, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T06:32:57.302519+00:00", "submitted_at": "2026-04-25T09:31:02.658752+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "460a30f8eec04cafbe0510dac8f38c54", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "d6ac858801c8209bf45f2a2a56dadaa26f957029bef685a65638b57da6510bab", "result_fingerprint": "4d09e876f354effe22a87bfc7455f2889618b02db81ba5c4b240c57188390148", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
23
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 74.04, "avg_power_w": 33.8, "max_power_w": 39.6, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1770.27, "p50_ttft_ms": 1373.96, "p99_ttft_ms": 6314.1, "avg_itl_ms": 408.4, "p50_itl_ms": 397.74, "p99_itl_ms": 965.59, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T06:50:44.318253+00:00", "submitted_at": "2026-04-25T09:31:02.660524+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "d5c80b4b8cd145d9bf41042ef610ad7e", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "32f3916064cbc9d104932b2447395247da778c0807647712478d74f0bef22e9f", "result_fingerprint": "e67daca82b3b8232b8ee32774cf89de9bd928d22211b13fb6d5c83c09c90dac8", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
24
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 54.17, "avg_power_w": 36.0, "max_power_w": 40.2, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1605.86, "p50_ttft_ms": 1448.97, "p99_ttft_ms": 5618.66, "avg_itl_ms": 282.46, "p50_itl_ms": 278.27, "p99_itl_ms": 503.28, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T07:02:58.469878+00:00", "submitted_at": "2026-04-25T09:31:02.662182+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "fc24bc5424204e71b33a03a3993d5010", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "2f661fb8e1d93fca9c927c5aea46501f29f10607a263c44cd358356506618739", "result_fingerprint": "7ba4b5ec565d387c907d3eeba6ab8fda80d719831a55d42b1bac93a61370bdb8", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
25
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 47.48, "avg_power_w": 35.6, "max_power_w": 39.6, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 880.28, "p50_ttft_ms": 641.67, "p99_ttft_ms": 2092.34, "avg_itl_ms": 163.04, "p50_itl_ms": 166.67, "p99_itl_ms": 199.85, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T07:10:04.135580+00:00", "submitted_at": "2026-04-25T09:31:02.663816+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "3b1da76a6fa942d6893ce6fffc844a37", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "74b03b6f6df22681913df17bf516260629985a9144b46d5a4ceebae0240752ea", "result_fingerprint": "9c874c4f2bd13a06954d8640187765ff4ca9d31bb17be1a8e1f014c53aa7cc72", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
26
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 43.8, "avg_power_w": 35.4, "max_power_w": 37.4, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 372.96, "p50_ttft_ms": 316.63, "p99_ttft_ms": 720.33, "avg_itl_ms": 88.95, "p50_itl_ms": 95.4, "p99_itl_ms": 118.1, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T07:13:49.279561+00:00", "submitted_at": "2026-04-25T09:31:02.665372+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "e855cdc2cff14ff28ddac698d35a5a62", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "b15e8a9c664351fb6a2b38961960e5dc4c47d79f67d3ccfb6dcb5c8f343d1557", "result_fingerprint": "ca84bdc194a0f3bf419dca6d02473a50add202b2d9d9bcec82888be31b957597", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
27
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 41.94, "avg_power_w": 34.7, "max_power_w": 37.1, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 230.5, "p50_ttft_ms": 198.55, "p99_ttft_ms": 510.63, "avg_itl_ms": 46.75, "p50_itl_ms": 49.77, "p99_itl_ms": 52.62, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T07:15:50.056137+00:00", "submitted_at": "2026-04-25T09:31:02.666886+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "e9ef4d3400a149aaa3ba24f371ef996a", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "905ed969c74d171fcc263314ea90687eeb71d196c207c879e320a9c6608b9cbe", "result_fingerprint": "309521b90c8bc000e9454e8c9eff6647193dd1122c8fa724e2cd29b4cafd9b36", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
28
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 41.14, "avg_power_w": 35.9, "max_power_w": 37.5, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 139.02, "p50_ttft_ms": 151.25, "p99_ttft_ms": 288.82, "avg_itl_ms": 23.84, "p50_itl_ms": 24.0, "p99_itl_ms": 25.84, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T07:16:56.681277+00:00", "submitted_at": "2026-04-25T09:31:02.668793+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "0443a4d8e4ff4fcfa65096dd13852d3e", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "3d832cfdd2267152bba74c298154eb86d1e1134c99a2fbe7834971cd4f1b7ab0", "result_fingerprint": "1a191d60df46292da29929cdc37d84d8e9373794f3a9c7f98af222908463bfea", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
29
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 74.81, "avg_power_w": 34.2, "max_power_w": 40.7, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1735.68, "p50_ttft_ms": 1433.98, "p99_ttft_ms": 6497.38, "avg_itl_ms": 406.36, "p50_itl_ms": 393.02, "p99_itl_ms": 956.43, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T07:34:17.526676+00:00", "submitted_at": "2026-04-25T09:31:02.670319+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "50227b09ed5d4cb390eb30f6c8e5dcc3", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "e36df304e7c93db6d542d6d790db474feed4e761b71a65ee7c718af020a8b186", "result_fingerprint": "1ee9b7d74c4101f00348cb6ae00c26399193c7c83b72c5c1d522cb5a83588c45", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
30
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 52.9, "avg_power_w": 36.5, "max_power_w": 40.5, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1577.58, "p50_ttft_ms": 1438.4, "p99_ttft_ms": 4646.68, "avg_itl_ms": 285.58, "p50_itl_ms": 282.32, "p99_itl_ms": 477.06, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T07:46:31.177960+00:00", "submitted_at": "2026-04-25T09:31:02.671791+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "510c3898626647bdb036c86df50c231b", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "7dd7796884b6a6ed5d2565166843ab50e2064eba7382e3eb93728312e7635e6a", "result_fingerprint": "b296f56388f98b264342fd85bdbfae6bb177e454d51149aca1c10b3fa12dc157", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
31
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 46.86, "avg_power_w": 36.8, "max_power_w": 39.3, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 659.11, "p50_ttft_ms": 504.43, "p99_ttft_ms": 1327.35, "avg_itl_ms": 163.57, "p50_itl_ms": 165.06, "p99_itl_ms": 232.58, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T07:53:18.184527+00:00", "submitted_at": "2026-04-25T09:31:02.673261+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "21392feda3f44309865fb885c80b658a", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "f3a0d53c7eb543341078b6c72b349ebe0d5ae75bda613d3ded15d42b624a45d3", "result_fingerprint": "04a951ed357dba1b291ebb017b06ff02497f67e952617928d2eb1bd624034d6a", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
32
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 41.77, "avg_power_w": 35.8, "max_power_w": 39.0, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 433.59, "p50_ttft_ms": 362.49, "p99_ttft_ms": 990.05, "avg_itl_ms": 93.43, "p50_itl_ms": 97.05, "p99_itl_ms": 117.74, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T07:57:11.937717+00:00", "submitted_at": "2026-04-25T09:31:02.674677+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "4d8c5981918542f69430e26fe9504558", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "719e7a5bc7501de2ec5e041d32fd78222902c70a540b825fa5f860f7174ae76f", "result_fingerprint": "ee2703adf8791cae0af0763b6e66c4db9fdd4d583d1370052714c825031529ff", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
33
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 41.54, "avg_power_w": 35.7, "max_power_w": 37.8, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 245.17, "p50_ttft_ms": 220.06, "p99_ttft_ms": 580.89, "avg_itl_ms": 46.98, "p50_itl_ms": 49.04, "p99_itl_ms": 51.25, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T07:59:18.878494+00:00", "submitted_at": "2026-04-25T09:31:02.676110+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "aac5782eccaf49f1bcc46c5a183d3ec8", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "cf8e1504fb3512a7a479e0b1bd8a51e6482d6cd7ca63128929b317b2f0e01d73", "result_fingerprint": "3def375cd4e9a781d520230536f407b632bf685a19e3575a227e70437b559ed6", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
34
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 32768, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 41.25, "avg_power_w": 36.3, "max_power_w": 37.8, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 131.98, "p50_ttft_ms": 149.76, "p99_ttft_ms": 272.53, "avg_itl_ms": 23.78, "p50_itl_ms": 23.88, "p99_itl_ms": 25.8, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T08:00:23.198614+00:00", "submitted_at": "2026-04-25T09:31:02.677494+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "19303cc695634992a5904c0545152e25", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "4965eb63287d84c24bb9c64c77f15b6969125a6e41e257bb2812cedd116301d6", "result_fingerprint": "58e13ffb89836e659fcdb3fae88b1dd37e09725250c3a03d059a3c963287f883", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
35
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 72.25, "avg_power_w": 32.6, "max_power_w": 42.5, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1499.03, "p50_ttft_ms": 1241.69, "p99_ttft_ms": 3982.93, "avg_itl_ms": 422.88, "p50_itl_ms": 410.86, "p99_itl_ms": 1005.31, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T08:18:38.499675+00:00", "submitted_at": "2026-04-25T09:31:02.678888+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "e55cc54b8d934414ac934210c425b3a1", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "2f73cb20d260f0c1a4b2126297b4d960050716025397da611d6d749da8cbbfac", "result_fingerprint": "49d451779f59da0d227f9900887728243fc35eddd1f6c9d53fa89df3d6891823", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
36
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 50.29, "avg_power_w": 35.4, "max_power_w": 39.6, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1470.8, "p50_ttft_ms": 1131.4, "p99_ttft_ms": 3691.54, "avg_itl_ms": 305.11, "p50_itl_ms": 304.84, "p99_itl_ms": 564.38, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T08:31:40.336987+00:00", "submitted_at": "2026-04-25T09:31:02.680210+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "24f0d699bdb54c5785a5f47d9043313f", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "245403869baabf04d4e0772a9e6534d9880b689522bb3afcf161c55d5342f98d", "result_fingerprint": "17d2e871b4a4c4393b0466ca2a41087be239fb4a727eeb1c87cc0086c6fab8f2", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
37
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 48.46, "avg_power_w": 35.3, "max_power_w": 39.0, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 699.12, "p50_ttft_ms": 558.17, "p99_ttft_ms": 1442.29, "avg_itl_ms": 160.51, "p50_itl_ms": 157.59, "p99_itl_ms": 232.17, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T08:38:36.216331+00:00", "submitted_at": "2026-04-25T09:31:02.681519+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "500ae706a00e4eb5a518a24fd93f397d", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "f7b852e50cdad09fc04f8d6a5689ae73310d4cb8f0fab3b29b721ad647c32032", "result_fingerprint": "ff118301b304dce93c8c445979d92f1cc5ffefddf39359b81ab536cae4ee9c04", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
38
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 44.47, "avg_power_w": 34.6, "max_power_w": 39.3, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 437.92, "p50_ttft_ms": 368.21, "p99_ttft_ms": 932.23, "avg_itl_ms": 87.07, "p50_itl_ms": 84.77, "p99_itl_ms": 119.45, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T08:42:19.523853+00:00", "submitted_at": "2026-04-25T09:31:02.682807+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "59908d4fa97647f99853d68e4b5b80a0", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "285fd2ef3c0346c4f93d295cf29be8790922ab9bf1988ec0d859e93f5c090d9c", "result_fingerprint": "23f68208c68f696e8a945808622493a5f2576c64f68330d3b569379c84798029", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
39
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 43.22, "avg_power_w": 35.5, "max_power_w": 38.8, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 215.45, "p50_ttft_ms": 163.25, "p99_ttft_ms": 389.51, "avg_itl_ms": 43.62, "p50_itl_ms": 48.97, "p99_itl_ms": 54.19, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T08:44:14.914547+00:00", "submitted_at": "2026-04-25T09:31:02.684053+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "147567942e874b8881be2543d0f7afb1", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "125784d9d6f39294cdda6635a9cbac4188e2344b92ed81960b4f353aa4cc57af", "result_fingerprint": "65f0992ca14df91fc579a60be502fb8d44e05fb0a47ef3804d6dba730bde90cc", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
40
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 512, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 40.68, "avg_power_w": 35.5, "max_power_w": 38.8, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 140.09, "p50_ttft_ms": 159.54, "p99_ttft_ms": 287.74, "avg_itl_ms": 24.06, "p50_itl_ms": 23.91, "p99_itl_ms": 26.58, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T08:45:17.224361+00:00", "submitted_at": "2026-04-25T09:31:02.685295+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "c8bf22e7f37f4a54bc0904c737509319", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "cb93db9a7f4a19b78338792030853a375513967d8ab89584ff911c3ec56460e9", "result_fingerprint": "b18d7c9e30b4917e24727d4463e9f54464ef187a3d86fc02e4cd34e8006abf96", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
41
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 32, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 320, "n_predict": 256, "throughput_tok_s": 69.46, "avg_power_w": 32.0, "max_power_w": 42.6, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 2275.1, "p50_ttft_ms": 1573.65, "p99_ttft_ms": 10949.16, "avg_itl_ms": 438.35, "p50_itl_ms": 424.79, "p99_itl_ms": 1052.06, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T09:04:05.798431+00:00", "submitted_at": "2026-04-25T09:31:02.686604+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "ffb66273d41a429baca2ce0119474167", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "f908182474b1c427655c058923c15d3db24c9ac3d58484a6aefc610e36d037d2", "result_fingerprint": "b2e4e8ac3aff146c3c33a914a555458af3e6504451190514ef262ad4d0fcc91b", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
42
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 16, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 160, "n_predict": 256, "throughput_tok_s": 52.23, "avg_power_w": 36.4, "max_power_w": 41.6, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 1392.64, "p50_ttft_ms": 1230.58, "p99_ttft_ms": 3844.63, "avg_itl_ms": 295.44, "p50_itl_ms": 293.3, "p99_itl_ms": 523.39, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T09:16:44.092003+00:00", "submitted_at": "2026-04-25T09:31:02.687861+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "4e75931160dd4fb1877979e04b278b6d", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "a28abd3e2ea55fdf77a77027c4812d5e4ddd747b271bb5994f51187bf42767a4", "result_fingerprint": "cb0d25e7f687c715f4bd1d98ea6afb65dee718d55f678f788558a373ad9f124e", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
43
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 8, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 80, "n_predict": 256, "throughput_tok_s": 47.03, "avg_power_w": 35.8, "max_power_w": 39.6, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 668.58, "p50_ttft_ms": 532.33, "p99_ttft_ms": 1661.17, "avg_itl_ms": 164.49, "p50_itl_ms": 166.74, "p99_itl_ms": 224.63, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T09:23:40.565975+00:00", "submitted_at": "2026-04-25T09:31:02.689105+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "a51f602ad303485da48a95f91d913d0c", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "dcdf1312f3687595b16e399fce0bb383b4dcf9499e8165a068b9c6081619b93b", "result_fingerprint": "1944f107af2cdb3b018432c58ca18a3040e51b5b9b49609d89c8b80edb11f783", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
44
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 4, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 40, "n_predict": 256, "throughput_tok_s": 43.51, "avg_power_w": 35.6, "max_power_w": 37.3, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 427.13, "p50_ttft_ms": 298.76, "p99_ttft_ms": 796.99, "avg_itl_ms": 88.24, "p50_itl_ms": 94.59, "p99_itl_ms": 107.37, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T09:27:22.264797+00:00", "submitted_at": "2026-04-25T09:31:02.690343+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "f166ddf3d0c44e8ea08db3481c09abf5", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "94579162d6e69c7ee8543172b1d41d9e5e60edfbb77c1dad3aa1c683585fc4c2", "result_fingerprint": "c3c2e864c3ba489f77c8f4ec2340a2b2f003cfcc543746d68aca86aa2c8f7412", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
45
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 2, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 20, "n_predict": 256, "throughput_tok_s": 40.84, "avg_power_w": 34.9, "max_power_w": 36.9, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 249.36, "p50_ttft_ms": 299.42, "p99_ttft_ms": 492.44, "avg_itl_ms": 48.19, "p50_itl_ms": 49.93, "p99_itl_ms": 50.42, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T09:29:33.178120+00:00", "submitted_at": "2026-04-25T09:31:02.691579+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "015d59a88f694549845161ad5f42fad5", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "dec6515d407239f9b3d12f489eb9b9431e38a629c74e1a44a51ead1976038c34", "result_fingerprint": "5e47cc1a44b793dfce1048e98ca3121aa2ee781b7f5e6599bebde33a36d33f9f", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}
46
+ {"model": "unsloth/Qwen3.6-35B-A3B-GGUF/Qwen3.6-35B-A3B-UD-Q5_K_S.gguf", "model_base": "Qwen3.6-35B-A3B-UD", "quant": "Q5_K_S", "model_org": "unsloth", "model_repo": "unsloth/Qwen3.6-35B-A3B-GGUF", "runner_type": "llama-server", "llm_engine_name": "llama.cpp", "llm_engine_version": null, "gpu_name": "Apple M4 Pro", "gpu_vram_gb": 64.0, "gpu_driver": "Metal 4", "gpu_count": 1, "gpu_names": "Apple M4 Pro", "gpu_total_vram_gb": 64.0, "backends": "Metal 4", "cpu_model": "Apple M4 Pro", "n_ctx": 130064, "n_batch": 1024, "split_mode": null, "tensor_split": null, "concurrent_users": 1, "task_type": "text-generation", "prompt_dataset": "sharegpt-v3", "num_prompts": 10, "n_predict": 256, "throughput_tok_s": 41.13, "avg_power_w": 35.8, "max_power_w": 37.7, "avg_gpu_temp_c": null, "max_gpu_temp_c": null, "avg_cpu_temp_c": null, "max_cpu_temp_c": null, "avg_fan_speed_rpm": null, "max_fan_speed_rpm": null, "avg_ttft_ms": 136.11, "p50_ttft_ms": 151.72, "p99_ttft_ms": 267.64, "avg_itl_ms": 23.84, "p50_itl_ms": 24.02, "p99_itl_ms": 25.62, "quality_score": null, "os_system": "Darwin", "os_release": "25.3.0", "os_machine": "arm64", "os_distro": "macOS", "os_distro_version": "26.3.1", "cpu_cores": "14 (10 Performance and 4 Efficiency)", "ram_total_gb": 64.0, "submitter": "Paul Lee", "timestamp": "2026-04-25T09:30:38.002020+00:00", "submitted_at": "2026-04-25T09:31:02.692803+00:00", "schema_version": "0.1.0", "benchmark_version": "0.1.0", "suite_run_id": "36c6b87566714f979aa9646454e11d68", "submission_id": "36c6b87566714f979aa9646454e11d68", "row_id": "a8fbc743f1c54b02bbb0d8a0c082bee6", "machine_fingerprint": "d1fb490248ff327581ae6b8f0bc7f566cae8855217831fdfd0fd68fc616b9a43", "run_fingerprint": "e97baaf856021f3c74b8cc8623809a1bf8536e8104d579b152c5b38262c84201", "result_fingerprint": "208c3e8327bfe124fbc2bde9a1a012363c424b1afa348bb12ddc80bad1397166", "source_file_sha256": "f21ad5e3ae13ea3bd48c37e23432ce7501ad6d31efcc7605e5da4cc2535c9309", "tags": null}