Instructions to use AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit 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("AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit")
config = load_config("AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit")

# 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

How to use AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-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 "AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit"

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": "AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

OpenClaw new

How to use AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-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 "AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-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 "AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-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"

Hermes Agent

How to use AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-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 "AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-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 AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit

Run Hermes

hermes

Atomic Chat

AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit

Development evidence AXQuant mixed-precision pack (not Tier-1 certified). No MTP claim.

Field	Value
Source	`Qwen/Qwen3-VL-30B-A3B-Instruct@9c4b90e1e4ba969fd3b5378b57d966d725f1b86c`
Product class	`4bit`
Measured total BPW	`4.860055`
Convert backend	MLX-VLM (affine RTN)
Primary runtime	AX Engine
Compatible runtime	MLX-VLM (vision / image-to-text)
Vision tower	Protected BF16
MTP	Not present / not claimed
Evidence	Architecture-prior plan + convert + mlx-vlm image smoke

Use with mlx-vlm

pip install -U mlx-vlm
python -m mlx_vlm.generate --model AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit \
  --max-tokens 100 --temperature 0.0 --prompt "Describe this image." --image <path.png>

Notes

Converted with axquant on Apple Silicon.
Sibling Thinking / other VL MoE sizes are not promoted by this pack.
AX Engine is the declared primary product runtime; install and load via your AX Engine workflow when available.

Modalities (capability-gated)

Text checkpoint Tier 1 does not imply vision or audio quality. Vision present=true on a pack is not a quality pass.

Modality	Claim	Supported	Reason
Vision	`smoke-certified`	`true`	vision runtime smoke passed on df-macstudio-m2 (mlx-vlm); quality suite not certified. Evidence: /Users/akiralam/code/axquant/docs/certifications/evidence/modality-recert-macstudio-m2/results/qwen3-vl-30b-axq4.json
Audio	`not-applicable`	`false`	audio not supported (no tower config and no sidecar weights)

Downloads last month: 47

Safetensors

Model size

6B params

Tensor type

BF16

U32

MLX

Hardware compatibility

4-bit

Model tree for AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit

Base model

Qwen/Qwen3-VL-30B-A3B-Instruct

Quantized

(60)

this model

Collections including AutomatosX/AX-Qwen3-VL-30B-A3B-Instruct-MLX-AXQ-4bit