Instructions to use Jundot/clef-flash-oQ2.5e with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Jundot/clef-flash-oQ2.5e 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-oQ2.5e")
config = load_config("Jundot/clef-flash-oQ2.5e")

# 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-oQ2.5e 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-oQ2.5e"

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-oQ2.5e"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use Jundot/clef-flash-oQ2.5e 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-oQ2.5e"

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-oQ2.5e

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Jundot/clef-flash-oQ2.5e 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-oQ2.5e"

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-oQ2.5e" \
  --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-oQ2.5e

This model was quantized using oQ (oMLX v0.7.1) mixed-precision quantization with Improved oQe Quantization (jundot/omlx#4385).

A mixed-precision MLX checkpoint of Cloudflare/clef-flash, a 9B decision model post-trained from Qwen/Qwen3.5-9B that turns a state and a schema of typed questions into decisions, returning a probability for every allowed option of every question in a single forward pass. It includes the Qwen3.5-9B 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.

The effective average is 4.011 bits/weight for the complete checkpoint.

Quantization and Bit Distribution

oQ2.5e allocates bits by measured layer sensitivity, then rounds each group with an importance matrix (128 samples x 512 tokens) and a weighted least-squares refit of its scale and bias.

Storage format	Logical weights	Share of total	Tensor storage	Effective bits/weight
Affine 2-bit, group 64	7.113B	74.624%	2.070 GiB	2.50
Affine 5-bit, group 64	0.768B	8.057%	0.492 GiB	5.50
Affine 6-bit, group 64	1.072B	11.246%	0.811 GiB	6.50
BF16	0.579B	6.073%	1.078 GiB	16.00
Total	9.532B	100%	4.451 GiB	4.011

Storage Breakdown

Component	Logical weights	Storage
Language backbone	8.954B	3.624 GB / 3.375 GiB
Vision encoder	0.456B	0.912 GB / 0.849 GiB
Joint schema head	0.122B	0.244 GB / 0.227 GiB
Total	9.532B	4.779 GB / 4.451 GiB

License

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

Downloads last month: 28

Safetensors

Model size

9B params

Tensor type

U32

BF16

MLX

Hardware compatibility

2-bit

Model tree for Jundot/clef-flash-oQ2.5e

Base model

Qwen/Qwen3.5-9B-Base

Finetuned

Qwen/Qwen3.5-9B

Finetuned

Cloudflare/clef-flash

Quantized

(54)

this model