Instructions to use deepsweet/Qwen3.6-27B-MLX-oQ8-FP16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use deepsweet/Qwen3.6-27B-MLX-oQ8-FP16 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("deepsweet/Qwen3.6-27B-MLX-oQ8-FP16")

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
LM Studio

Pi new

How to use deepsweet/Qwen3.6-27B-MLX-oQ8-FP16 with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "deepsweet/Qwen3.6-27B-MLX-oQ8-FP16"

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "deepsweet/Qwen3.6-27B-MLX-oQ8-FP16"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use deepsweet/Qwen3.6-27B-MLX-oQ8-FP16 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 "deepsweet/Qwen3.6-27B-MLX-oQ8-FP16"

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 deepsweet/Qwen3.6-27B-MLX-oQ8-FP16

Run Hermes

hermes

MLX LM

How to use deepsweet/Qwen3.6-27B-MLX-oQ8-FP16 with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "deepsweet/Qwen3.6-27B-MLX-oQ8-FP16"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "deepsweet/Qwen3.6-27B-MLX-oQ8-FP16"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "deepsweet/Qwen3.6-27B-MLX-oQ8-FP16",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

This model was converted to MLX format and quantized from Qwen3.6-27B using oMLX.

What is "oQ"?

See "oQ: oMLX Universal Dynamic Quantization" for details.

Quantizations

Unlike the Qwen3.6-35B-A3B collection, work on the full Qwen3.6-27B 4x4 table is still ongoing. So if you see a 404 link below, feel free to ask me to prioritize it.

Text-Only	Vision-Language	Text-Only FP16	Vision-Language FP16
MLX-oQ8	MLX-VL-oQ8	MLX-oQ8-FP16	MLX-VL-oQ8-FP16
MLX-oQ6	MLX-VL-oQ6	MLX-oQ6-FP16	MLX-VL-oQ6-FP16
MLX-oQ5	MLX-VL-oQ5	MLX-oQ5-FP16	MLX-VL-oQ5-FP16
MLX-oQ4	MLX-VL-oQ4	MLX-oQ4-FP16	MLX-VL-oQ4-FP16

See "Evaluation of various MLX quantizations" for details:

What is "VL"?

"VL" is Vision-Language, meaning quantization preserves the original model's multimodality.

No "VL" means quantization is Text-Only.

What is "FP16"?

"FP16" is an M1/M2 Apple Silicon tweak that delivers a very noticeable prompt processing boost, because older M-series lack native BF16 hardware support. See "Metal FP32 Vs BF16 Vs FP16 benchmark" for details.

No "FP16" means quantization is better suited for M3+ Apple Silicon.

Downloads last month: 4,218

Safetensors

Model size

8B params

Tensor type

F16

U32

MLX

Hardware compatibility

8-bit

Model tree for deepsweet/Qwen3.6-27B-MLX-oQ8-FP16

Base model

Qwen/Qwen3.6-27B

Quantized

(360)

this model

Collection including deepsweet/Qwen3.6-27B-MLX-oQ8-FP16

Qwen3.6-27B

Collection

10 items • Updated 6 days ago • 3