Instructions to use Jundot/clef-flash-oQ4e with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Jundot/clef-flash-oQ4e 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("Jundot/clef-flash-oQ4e")
config = load_config("Jundot/clef-flash-oQ4e")

# 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 Jundot/clef-flash-oQ4e with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "Jundot/clef-flash-oQ4e"

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": "Jundot/clef-flash-oQ4e"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use Jundot/clef-flash-oQ4e 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 "Jundot/clef-flash-oQ4e"

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 Jundot/clef-flash-oQ4e

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Jundot/clef-flash-oQ4e with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "Jundot/clef-flash-oQ4e"

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 "Jundot/clef-flash-oQ4e" \
  --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"

clef-flash-oQ4e

This model was quantized using oQ mixed-precision quantization.

A mixed-precision MLX checkpoint of Cloudflare/clef-flash, a decision model that answers typed questions (noul, choice, score) about text and images with a probability for each option. It includes the Qwen3.5 language backbone, the vision encoder and the joint schema head.

The joint schema head (joint_head.safetensors, joint_head_config.json) is copied unchanged from the source. oMLX serves this checkpoint through POST /v1/systemone once decision model support (jundot/omlx#4315) is available.

The effective average is 5.574 bits/weight for the complete checkpoint.

Quantization and Bit Distribution

4-bit affine quantization, group size 64, for most language-backbone projections and the token embeddings.
5-bit affine quantization, group size 64, for projections boosted by measured layer sensitivity.
BF16 for the vision encoder, the joint schema head, norms and the remaining small tensors.

The oQ4e allocation uses measured layer sensitivity and importance-matrix calibration (128 samples x 512 tokens, built-in code and multilingual set).

Storage format	Logical weights	Share of total	Tensor storage	Effective bits/weight
Affine 4-bit, group 64	8.288B	84.469%	4.342 GiB	4.50
Affine 5-bit, group 64	0.665B	6.780%	0.426 GiB	5.50
BF16	0.859B	8.751%	1.599 GiB	16.00
Total	9.811B	100%	6.367 GiB	5.574

Storage Breakdown

Component	Logical weights	Storage
Language backbone	9.234B	5.681 GB / 5.291 GiB
Vision encoder	0.456B	0.912 GB / 0.849 GiB
Joint schema head	0.122B	0.244 GB / 0.227 GiB
Total	9.811B	6.837 GB / 6.367 GiB

License

Apache-2.0, following the source model. See LICENSE.

Downloads last month: 57

Safetensors

Model size

9B params

Tensor type

U32

BF16

MLX

Hardware compatibility

4-bit

Model tree for Jundot/clef-flash-oQ4e

Base model

Qwen/Qwen3.5-9B-Base

Finetuned

Qwen/Qwen3.5-9B

Finetuned

Cloudflare/clef-flash

Quantized

(35)

this model