Instructions to use spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit 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("spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit")

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

How to use spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit"

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": "spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit 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 "spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit"

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 spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit"

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 "spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit" \
  --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"

spicyneuron commited on Apr 16

Commit

1a0dcd0

verified ·

1 Parent(s): a4144b9

Update README.md

Browse files

Files changed (1) hide show

README.md +37 -3

README.md CHANGED Viewed

@@ -9,8 +9,12 @@ base_model: Qwen/Qwen3.6-35B-A3B
 ---
 [Qwen3.6-35B-A3B](https://huggingface.co/Qwen/Qwen3.6-35B-A3B) optimized for MLX.
-This quant does not support image input.
 # Usage
@@ -19,9 +23,40 @@ This quant does not support image input.
 uvx --from mlx-lm mlx_lm.server \
   --host 127.0.0.1 \
   --port 8080 \
-  --model spicyneuron/Qwen3.6-35B-A3B-MLX-5.4bit
 ```
 # Methodology
 Quantized with a [mlx-lm fork](https://github.com/ml-explore/mlx-lm/pull/922), drawing inspiration from Unsloth/AesSedai/ubergarm style mixed-precision GGUFs.
@@ -29,4 +64,3 @@ MLX quantization options differ than llama.cpp, but the principles are the same:
 - Sensitive layers like MoE routing, attention, and output embeddings get higher precision
 - More tolerant layers like MoE experts get lower precision

 ---
 [Qwen3.6-35B-A3B](https://huggingface.co/Qwen/Qwen3.6-35B-A3B) optimized for MLX.
+- 4-bit baseline with important layers at 8-bit and BF16.
+- This quant does not support image input.
+I ended up selecting two winners from my trials. This is the speed+ version, and
+here's the [quality+](https://huggingface.co/spicyneuron/Qwen3.6-35B-A3B-MLX-5.4bit) version.
 # Usage
 uvx --from mlx-lm mlx_lm.server \
   --host 127.0.0.1 \
   --port 8080 \
+  --model spicyneuron/Qwen3.6-35B-A3B-MLX-4.8bit
+```
+# Benchmarks
+metric | mlx-community/ Qwen3.6-35B-A3B-4bit | mlx-community/ Qwen3.6-35B-A3B-4.4bit-msq | 4.8 bit (this model) | 5.4 bit
+--- | --- | --- | --- | ---
+bpw | 4.503 | 4.787 | 4.788 | 5.438
+peak memory (1024/512) | 20.683 | 21.922 | 21.928 | 24.741
+prompt tok/s (1024) | 2719.4470 ± 15.2250 | 2695.9370 ± 12.5260 | 2734.5260 ± 3.8810 | 2665.3060 ± 11.4520
+gen tok/s (512) | 108.4990 ± 0.4910 | 94.2940 ± 0.3650 | 97.2820 ± 0.0800 | 89.4920 ± 0.2610
+kl divergence | 0.0838 ± 0.0008 | 0.1689 ± 0.0015 | 0.0244 ± 0.0004 | 0.0189 ± 0.0003
+perplexity | 4.6150 ± 0.0320 | 4.2490 ± 0.0280 | 4.6410 ± 0.0320 | 4.6440 ± 0.0320
+hellaswag | 0.5560 ± 0.0220 | 0.5780 ± 0.0220 | 0.5440 ± 0.0220 | 0.5370 ± 0.0110
+piqa | 0.7940 ± 0.0180 | 0.7920 ± 0.0180 | 0.7920 ± 0.0180 | 0.7980 ± 0.0180
+winogrande | 0.7260 ± 0.0200 | 0.7400 ± 0.0200 | 0.7120 ± 0.0200 | 0.7100 ± 0.0200
+I've moved over to using speed + KL divergence as my primary optimization metrics.
+Hellaswag, PIQA, Winogrande, and perplexity are kept as sanity checks, though these require
+high sample sizes to get usable signal.
+Tested on a Mac Studio M3 Ultra with:
+```
+mlx_lm.convert --hf-path Qwen/Qwen3.6-35B-A3B --mlx-path ./mlx && mlx_lm.kld --baseline-model ./mlx
+mlx_lm.perplexity --sequence-length 512 --seed 123
+mlx_lm.benchmark --prompt-tokens 1024 --generation-tokens 512 --num-trials 5
+mlx_lm.evaluate --tasks hellaswag --seed 123 --num-shots 0 --limit 500
+mlx_lm.evaluate --tasks piqa --seed 123 --num-shots 0 --limit 500
+mlx_lm.evaluate --tasks winogrande --seed 123 --num-shots 0 --limit 500
 ```
+`mlx_lm.kld` is still an [open PR](https://github.com/ml-explore/mlx-lm/pull/1146).
 # Methodology
 Quantized with a [mlx-lm fork](https://github.com/ml-explore/mlx-lm/pull/922), drawing inspiration from Unsloth/AesSedai/ubergarm style mixed-precision GGUFs.
 - Sensitive layers like MoE routing, attention, and output embeddings get higher precision
 - More tolerant layers like MoE experts get lower precision