Image-Text-to-Text
MLX
Safetensors
qwen3_5
mtplx
qwen3.6
speculative-decoding
multi-token-prediction
5-bit
conversational
Instructions to use deanzchen/Qwen3.6-27B-Q5-MTPLX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use deanzchen/Qwen3.6-27B-Q5-MTPLX with MLX:
# Make sure mlx-vlm is installed # pip install --upgrade mlx-vlm from mlx_vlm import load, generate from mlx_vlm.prompt_utils import apply_chat_template from mlx_vlm.utils import load_config # Load the model model, processor = load("deanzchen/Qwen3.6-27B-Q5-MTPLX") config = load_config("deanzchen/Qwen3.6-27B-Q5-MTPLX") # Prepare input image = ["http://images.cocodataset.org/val2017/000000039769.jpg"] prompt = "Describe this image." # Apply chat template formatted_prompt = apply_chat_template( processor, config, prompt, num_images=1 ) # Generate output output = generate(model, processor, formatted_prompt, image) print(output) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Pi
How to use deanzchen/Qwen3.6-27B-Q5-MTPLX with Pi:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "deanzchen/Qwen3.6-27B-Q5-MTPLX"
Configure the model in Pi
# Install Pi: npm install -g @earendil-works/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "mlx-lm": { "baseUrl": "http://localhost:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "deanzchen/Qwen3.6-27B-Q5-MTPLX" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Hermes Agent
How to use deanzchen/Qwen3.6-27B-Q5-MTPLX with Hermes Agent:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "deanzchen/Qwen3.6-27B-Q5-MTPLX"
Configure Hermes
# Install Hermes: curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash hermes setup # Point Hermes at the local server: hermes config set model.provider custom hermes config set model.base_url http://127.0.0.1:8080/v1 hermes config set model.default deanzchen/Qwen3.6-27B-Q5-MTPLX
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use deanzchen/Qwen3.6-27B-Q5-MTPLX with OpenClaw:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "deanzchen/Qwen3.6-27B-Q5-MTPLX"
Configure OpenClaw
# Install OpenClaw: npm install -g openclaw@latest # Register the local server and set it as the default model: openclaw onboard --non-interactive --mode local \ --auth-choice custom-api-key \ --custom-base-url http://127.0.0.1:8080/v1 \ --custom-model-id "deanzchen/Qwen3.6-27B-Q5-MTPLX" \ --custom-provider-id mlx-lm \ --custom-compatibility openai \ --custom-text-input \ --accept-risk \ --skip-health
Run OpenClaw
openclaw agent --local --agent main --message "Hello from Hugging Face"
File size: 4,565 Bytes
7cd083c f007c89 7cd083c c1a80cd 7cd083c a6a0651 7cd083c f007c89 a6a0651 f007c89 7cd083c f007c89 cb349de 3f6a8b7 c1a80cd 7cd083c cb349de 7cd083c | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 | {
"schema_version": 1,
"artifact": "deanzchen/Qwen3.6-27B-Q5-MTPLX",
"hardware": {
"chip": "Apple M5 Max",
"gpu_cores": 40,
"unified_memory_gib": 48,
"power_mode_for_latest_check": "high_power"
},
"runtime": {
"mtplx": "2.0.2",
"mlx": "0.31.2",
"mlx_lm": "0.31.3",
"profile": "sustained",
"generation_mode": "mtp",
"depth": 2,
"context_window": 262144,
"prefill_chunk_tokens": 512,
"trunk_quantization": "5-bit affine group64",
"draft_head_quantization": "4-bit affine group64",
"target_sampler": {
"temperature": 0.6,
"top_p": 0.95,
"top_k": 20
},
"draft_sampler_for_ui_measurements": {
"temperature": 0.6,
"top_p": 0.95,
"top_k": 20
},
"paged_kv_quantization": "off",
"turboquant_note": "K8/V4 affects extreme-context cache capacity, not these short decode measurements."
},
"verification_schedule": {
"defer_verify_hidden_eval": true,
"verify_hidden_mode": "logits_first_committed_slice",
"exact_token_identity_verified": true
},
"expected_decode_tok_s": {
"direct_tuner_d2": {
"expected_range": [45.0, 50.0],
"warning_below": 45.0,
"failure_below": 42.0,
"workload": "mtplx long_code, thinking off, seed 0, 128 or 256 generated tokens"
},
"direct_tuner_d2_high_power": {
"expected_range": [45.0, 50.0],
"warning_below": 45.0,
"workload": "mtplx long_code, thinking off, seed 0, 128 generated tokens"
},
"openai_api_code_first": {
"expected_range": [40.0, 44.0],
"requirement": "Send X-MTPLX-Allow-Client-Controls: 1 with enable_thinking=false."
},
"eight_token_smoke": {
"expected_range": [35.0, 46.0],
"regression_gate": false,
"note": "Too short for stable throughput measurement."
}
},
"measurements": {
"historical_setup_128_d2": 48.799728436870346,
"historical_comparison_256_d2": 49.90875607132132,
"hub_direct_128_d2": 43.92720981573101,
"hub_high_power_direct_128_d2": 45.419032271264705,
"hub_post_reboot_high_power_direct_128_d2": 46.27235162628437,
"hub_post_reboot_high_power_direct_128_d2_variance_5x": {
"runs": [
46.47503852458389,
46.38583666459772,
46.33124692797028,
46.27492432091384,
46.316507149484536
],
"mean": 46.356710717510055,
"median": 46.33124692797028,
"sample_standard_deviation": 0.07714197438579953,
"coefficient_of_variation_percent": 0.16640950833610618,
"minimum": 46.27492432091384,
"maximum": 46.47503852458389,
"range": 0.20011420367004717,
"all_runs_above_45": true,
"acceptance_rate_all_runs": 1.0
},
"original_local_deferred_hidden_direct_128_d2_variance_5x": {
"runs": [
50.67376896277482,
50.9484927857385,
50.75530318826678,
50.7859195241367,
50.750990759454915
],
"mean": 50.782895044074344,
"median": 50.75530318826678,
"sample_standard_deviation": 0.10138966296877788,
"coefficient_of_variation_percent": 0.19965317629249388,
"minimum": 50.67376896277482,
"maximum": 50.9484927857385,
"all_runs_at_or_above_48_8": true,
"acceptance_rate_all_runs": 1.0,
"token_sha256": "f875ce74f006ba502dac263111e6c9b1b1b4bca03648c43f670c8cae8e0a58af"
},
"persisted_tune_wrapper_final_gate": {
"decode_tok_s": 49.81852161488856,
"acceptance_rate": 1.0,
"accepted_by_depth": [43, 42],
"drafted_by_depth": [43, 42],
"token_sha256": "f875ce74f006ba502dac263111e6c9b1b1b4bca03648c43f670c8cae8e0a58af",
"text_sha256": "abc2ec0ce79cbf90e38043eeacd0fc9d34d624780e6cc630b3d5fbd0e3ced05e",
"target_48_8_met": true
},
"hub_direct_256_d2_runs": [43.830748328265656, 43.695076417584225],
"hub_api_code_first_three_run_mean": 41.46785703207608,
"hub_api_auto_reasoning_three_run_mean": 39.68670477257572,
"latest_ar_128": 24.149735339889887,
"latest_d2_speedup_vs_ar": 1.916060402941681,
"high_power_improvement_percent": 5.338699681566217,
"post_reboot_improvement_percent": 1.878770445665201
},
"validation": {
"status": "pass",
"model_files_match_original_sha256": true,
"direct_128_accepted_by_depth": [43, 42],
"direct_128_drafted_by_depth": [43, 42],
"direct_256_acceptance_rate": 0.9883720930232558,
"conclusion": "No artifact or MTP acceptance regression. Current throughput is within the expected machine-state range."
}
}
|