Instructions to use nightmedia/Qwen3-Coder-Next-mxfp4-mlx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nightmedia/Qwen3-Coder-Next-mxfp4-mlx 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("nightmedia/Qwen3-Coder-Next-mxfp4-mlx")

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 nightmedia/Qwen3-Coder-Next-mxfp4-mlx with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "nightmedia/Qwen3-Coder-Next-mxfp4-mlx"

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": "nightmedia/Qwen3-Coder-Next-mxfp4-mlx"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use nightmedia/Qwen3-Coder-Next-mxfp4-mlx with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "nightmedia/Qwen3-Coder-Next-mxfp4-mlx"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "nightmedia/Qwen3-Coder-Next-mxfp4-mlx"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "nightmedia/Qwen3-Coder-Next-mxfp4-mlx",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use nightmedia/Qwen3-Coder-Next-mxfp4-mlx 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 "nightmedia/Qwen3-Coder-Next-mxfp4-mlx"

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 nightmedia/Qwen3-Coder-Next-mxfp4-mlx

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use nightmedia/Qwen3-Coder-Next-mxfp4-mlx with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "nightmedia/Qwen3-Coder-Next-mxfp4-mlx"

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 "nightmedia/Qwen3-Coder-Next-mxfp4-mlx" \
  --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"

Issue with Roo-Code Native Tool Calling: "XML tool calls are no longer supported" (Works with lmstudio-community version)

by cracosansan - opened Feb 10

Discussion

cracosansan

Feb 10

Hello,
I am encountering a specific issue with this model (nightmedia/Qwen3-Coder-Next-mxfp4-mlx) when using it with VS Code Roo-Code extension (v3.47.3) via a local LM Studio server (OpenAI Proxy).
The Issue:
When using Roo-Code in "Code" mode (which relies on function calling), the model fails with the following error from the extension:
code
Text
Date/time: 2026-02-10T07:12:55.394Z
Extension version: 3.47.3
Provider: openai (proxy)
Model: nightmedia/qwen3-coder-next-mlx

XML tool calls are no longer supported. Remove any XML tool markup (e.g. ...) and use native tool calling instead.
The "Ask" mode works fine, presumably because it doesn't trigger tool usage heavily.
Comparison:
I have tested the exact same setup (same LM Studio version, same Jinja configuration, same System Prompt) with the lmstudio-community/qwen3-coder-next-mlx version of the model, and it works perfectly without this error.
Observation:
It seems this specific quantization/conversion (mxfp4-mlx) might be causing the model to output XML-style tool tags (e.g., ) instead of adhering to the OpenAI-compatible Native Tool Calling (JSON format) required by the latest versions of Roo-Code.
Could this be related to a missing or different chat_template configuration in the tokenizer_config.json compared to the community version? Or is it a side effect of the mxfp4 quantization affecting adherence to system prompts regarding tool formats?
Any insights or fixes would be appreciated. Thanks!

nightmedia

Owner Feb 10

•

edited Feb 10

Last time I updated the model, I changed the “tool_parser_type": "qwen3_coder",in the tokenizer config. I made the change in place—will double check and compare to see what’s missing

Try to comment out the chat_template in the config file and see if that makes a difference

This can be because I used the same template as the 4Bs and there might be some differences there

nightmedia

Owner Feb 10

Yeah, I reviewed and compared to the original, the only thing that is new is the "tool_parser_type": "qwen3_coder".

I added this at someone's suggestion that might solve some calls, try commenting that one out.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment