Instructions to use wang-yang/Qwen3.8-27B-MTPLX-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use wang-yang/Qwen3.8-27B-MTPLX-4bit with MLX:
# Make sure mlx-lm is installed # pip install --upgrade mlx-lm # Generate text with mlx-lm from mlx_lm import load, generate model, tokenizer = load("wang-yang/Qwen3.8-27B-MTPLX-4bit") prompt = "Write a story about Einstein" messages = [{"role": "user", "content": prompt}] prompt = tokenizer.apply_chat_template( messages, add_generation_prompt=True ) text = generate(model, tokenizer, prompt=prompt, verbose=True) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Pi
How to use wang-yang/Qwen3.8-27B-MTPLX-4bit with Pi:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "wang-yang/Qwen3.8-27B-MTPLX-4bit"
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": "wang-yang/Qwen3.8-27B-MTPLX-4bit" } ] } } }Run Pi
# Start Pi in your project directory: pi
- MLX LM
How to use wang-yang/Qwen3.8-27B-MTPLX-4bit with MLX LM:
Generate or start a chat session
# Install MLX LM uv tool install mlx-lm # Interactive chat REPL mlx_lm.chat --model "wang-yang/Qwen3.8-27B-MTPLX-4bit"
Run an OpenAI-compatible server
# Install MLX LM uv tool install mlx-lm # Start the server mlx_lm.server --model "wang-yang/Qwen3.8-27B-MTPLX-4bit" # Calling the OpenAI-compatible server with curl curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "wang-yang/Qwen3.8-27B-MTPLX-4bit", "messages": [ {"role": "user", "content": "Hello"} ] }' - Hermes Agent
How to use wang-yang/Qwen3.8-27B-MTPLX-4bit 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 "wang-yang/Qwen3.8-27B-MTPLX-4bit"
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 wang-yang/Qwen3.8-27B-MTPLX-4bit
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use wang-yang/Qwen3.8-27B-MTPLX-4bit with OpenClaw:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "wang-yang/Qwen3.8-27B-MTPLX-4bit"
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 "wang-yang/Qwen3.8-27B-MTPLX-4bit" \ --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"
| { | |
| "arch_id": "qwen3-next-mtp", | |
| "artifact_role": "forge-local", | |
| "base_trunk": "/Users/yang.wang/models/qwen3.8-27b-src", | |
| "exactness_baseline": {}, | |
| "forge_provenance": { | |
| "forge_inputs": { | |
| "mtp_source_path": "/Users/yang.wang/models/qwen3.8-27b-src", | |
| "trunk_path": "/Users/yang.wang/Documents/MTPLX/models/Qwen3.8-27B-MTPLX-4bit" | |
| }, | |
| "forge_recipe": { | |
| "body_bits": 4, | |
| "body_group_size": 64, | |
| "body_mode": "affine", | |
| "mtp_policy": "keep_bf16" | |
| }, | |
| "forged_at": "2026-08-24T00:28:10+09:00", | |
| "forged_locally": true, | |
| "mtp_contract": { | |
| "base_hidden_variant": "post_norm", | |
| "concat_order": "embedding_hidden", | |
| "hidden_variant": "post_norm", | |
| "mtp_position_mode": "local", | |
| "mtp_quant_group_size": 64, | |
| "mtp_quant_mode": "affine" | |
| }, | |
| "mtplx_version": "2.9.1", | |
| "published_to_hf": null, | |
| "source_format": "bf16_native", | |
| "source_repo": "/Users/yang.wang/models/qwen3.8-27b-src", | |
| "source_sha": null | |
| }, | |
| "mtp_contract": { | |
| "base_hidden_variant": "post_norm", | |
| "concat_order": "embedding_hidden", | |
| "hidden_variant": "post_norm", | |
| "mtp_position_mode": "local", | |
| "mtp_quant_group_size": 64, | |
| "mtp_quant_mode": "affine" | |
| }, | |
| "mtp_depth_max": 3, | |
| "mtp_sidecar": "mtp.safetensors", | |
| "mtplx_version": "2.9.1", | |
| "recommended_profile": "sustained", | |
| "sampler": { | |
| "temperature": 0.6, | |
| "top_k": 20, | |
| "top_p": 0.95 | |
| }, | |
| "speed_evidence": { | |
| "acceptance_by_depth": [ | |
| 1.0, | |
| 1.0, | |
| 0.8 | |
| ], | |
| "acceptance_collapsed": [], | |
| "artifact_fingerprint": "sha256:caa118baa7455f6e83aba9079cbba16ed04c375ffd5b987d9f45aecac19a6e6a", | |
| "depth": 3, | |
| "failure_reasons": [], | |
| "forge_verify_rows": [ | |
| { | |
| "acceptance_by_position": [], | |
| "depth": 0, | |
| "finish_reasons": { | |
| "length": 1 | |
| }, | |
| "hit_token_budget": true, | |
| "hit_token_budget_count": 1, | |
| "multiplier_vs_ar": 1.0, | |
| "quality_passed": true, | |
| "tok_s": 21.744767277204435, | |
| "verify_time_s": 1.9631269170204177 | |
| }, | |
| { | |
| "acceptance_by_position": [ | |
| 1.0 | |
| ], | |
| "depth": 1, | |
| "finish_reasons": { | |
| "length": 1 | |
| }, | |
| "hit_token_budget": true, | |
| "hit_token_budget_count": 1, | |
| "multiplier_vs_ar": 1.7178727553803572, | |
| "quality_passed": true, | |
| "tok_s": 37.35474327759581, | |
| "verify_time_s": 0.8072494564112276 | |
| }, | |
| { | |
| "acceptance_by_position": [ | |
| 1.0, | |
| 1.0 | |
| ], | |
| "depth": 2, | |
| "finish_reasons": { | |
| "length": 1 | |
| }, | |
| "hit_token_budget": true, | |
| "hit_token_budget_count": 1, | |
| "multiplier_vs_ar": 1.7711165965234528, | |
| "quality_passed": true, | |
| "tok_s": 38.512518212196866, | |
| "verify_time_s": 0.7674681238131598 | |
| }, | |
| { | |
| "acceptance_by_position": [ | |
| 1.0, | |
| 1.0, | |
| 0.8 | |
| ], | |
| "depth": 3, | |
| "finish_reasons": { | |
| "length": 1 | |
| }, | |
| "hit_token_budget": true, | |
| "hit_token_budget_count": 1, | |
| "multiplier_vs_ar": 2.295863870102533, | |
| "quality_passed": true, | |
| "tok_s": 49.92302555552149, | |
| "verify_time_s": 0.557183044264093 | |
| } | |
| ], | |
| "greedy_diagnostic": { | |
| "tok_s": 21.744767277204435 | |
| }, | |
| "quality_rejected": [], | |
| "tok_s": [ | |
| 49.92302555552149 | |
| ], | |
| "verdict": "mtp_depth_wins" | |
| }, | |
| "verified_on": { | |
| "hardware": "macOS-15.7.3-arm64-arm-64bit", | |
| "machine_arch": "arm64", | |
| "macos": "15.7.3", | |
| "model": "Qwen3.8-27B-MTPLX-4bit", | |
| "timestamp": "2026-08-24T00:28:10+09:00" | |
| }, | |
| "vision": { | |
| "prefix": "vision_tower.", | |
| "shards": [ | |
| "model-vision.safetensors" | |
| ], | |
| "tensor_count": 333 | |
| }, | |
| "mtp_file": "mtp.safetensors", | |
| "remaining_q4_to_q3": { | |
| "selected_leaf_types": [ | |
| "in_proj_z" | |
| ], | |
| "selected_layers": [ | |
| 2, | |
| 4, | |
| 5, | |
| 9, | |
| 10, | |
| 16, | |
| 17, | |
| 20, | |
| 21, | |
| 22, | |
| 24, | |
| 25, | |
| 29, | |
| 30, | |
| 32, | |
| 34, | |
| 36, | |
| 37, | |
| 38, | |
| 40, | |
| 41, | |
| 44, | |
| 45, | |
| 48, | |
| 49, | |
| 50, | |
| 53, | |
| 56, | |
| 60, | |
| 61, | |
| 62 | |
| ], | |
| "selected_module_count": 31, | |
| "group_size": 64, | |
| "mtp_sidecar_unchanged": true | |
| } | |
| } | |