Instructions to use openjev/OpenJev-Flash-9B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use openjev/OpenJev-Flash-9B-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M

Use Docker

docker model run hf.co/openjev/OpenJev-Flash-9B-GGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use openjev/OpenJev-Flash-9B-GGUF with Ollama:
```
ollama run hf.co/openjev/OpenJev-Flash-9B-GGUF:Q4_K_M
```
Unsloth Desktop

How to use openjev/OpenJev-Flash-9B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "openjev/OpenJev-Flash-9B-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use openjev/OpenJev-Flash-9B-GGUF with Docker Model Runner:
```
docker model run hf.co/openjev/OpenJev-Flash-9B-GGUF:Q4_K_M
```

Lemonade

How to use openjev/OpenJev-Flash-9B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull openjev/OpenJev-Flash-9B-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.OpenJev-Flash-9B-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use openjev/OpenJev-Flash-9B-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M

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 openjev/OpenJev-Flash-9B-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use openjev/OpenJev-Flash-9B-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf openjev/OpenJev-Flash-9B-GGUF:Q4_K_M

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 "openjev/OpenJev-Flash-9B-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

OpenJev Flash 9B, GGUF

Run OpenJev Flash 9B on a laptop, a consumer GPU or a CPU with llama.cpp. Four quantizations from 5.6 GB to 9.5 GB, one forward pass per decision.

On par with Cloudflare's Clef-Flash and ahead of Kev-9B and Nimble 9B on JevBench. The results, the API and the use cases are on the main card.

Q8_0 keeps the full model's accuracy: 79.7% against 80.0% for the 16-bit weights on a fixed 1,789-question check, the same answer on 99.0% of the questions.
Q4_K_M fits an 8 GB GPU or a 16 GB Mac at 5.6 GB and scores 79.3% on the same check.
Same prompt, same one-token answer as the main model: state, question, lettered options; the answer is the letter.

file	bits	size	fits
OpenJev-Flash-9B-Q4_K_M.gguf	~4.8	5.6 GB	8 GB GPUs, 16 GB Macs
OpenJev-Flash-9B-Q5_K_M.gguf	~5.7	6.5 GB	10 GB GPUs, 16 GB Macs
OpenJev-Flash-9B-Q6_K.gguf	~6.6	7.4 GB	10 GB GPUs, 16 GB Macs
OpenJev-Flash-9B-Q8_0.gguf	8.5	9.5 GB	12 GB GPUs, 24 GB Macs; closest to the 16-bit weights

Built from the 16-bit weights (openjev/OpenJev-Flash-9B, tag v1) with llama.cpp build b11147. Text input. SHA256SUMS and MANIFEST.json carry every hash and the validation numbers.

Run it

hf download openjev/OpenJev-Flash-9B-GGUF OpenJev-Flash-9B-Q4_K_M.gguf --local-dir .
llama-server -m OpenJev-Flash-9B-Q4_K_M.gguf -ngl 999 -c 16384 -np 2 --port 8080

Ask for a decision the way OpenJev was trained: state, question, lettered options, and read the letter.

curl -s localhost:8080/v1/chat/completions -H 'Content-Type: application/json' -d '{
  "messages": [{"role": "user", "content": "State:\nA checkout page shows: Subtotal $40, Shipping $5, a Place order button, and a Coupon field.\n\nQuestion: Which action completes the purchase?\nOptions:\n[A] click_coupon: type a coupon\n[B] click_place_order: click Place order\n[C] go_back: return to cart\n\nAnswer with the letter of the best option only."}],
  "max_tokens": 4, "temperature": 0, "chat_template_kwargs": {"enable_thinking": false}}'

For a probability per option, call /completion with "n_predict": 1 and "n_probs": 64 and read the letter tokens at the first output position.

For the typed /v1/systemone API with calibrated probabilities for every option, run the OpenJev helper (helper/shim.py in the main repository) as in the main card's Quick start, with the 9B settings READOUT_T=1.07 READOUT_NOUL_T=1.074766 READOUT_NOUL_BIAS=0 READOUT_TARGETED=1 READOUT_INSTR_STYLE=pyrepr; it serves on vLLM and on MLX.

Validation

The same 1,789 development questions and option letterings for every build, read the same way (letter probabilities at the first output position); validation/ has the per-question records.

build	accuracy	same answer as 16-bit
16-bit (vLLM)	80.0%	—
Q8_0 (llama.cpp)	79.7%	99.0%
Q4_K_M (llama.cpp)	79.3%	95.1%

Licence

Weights: CC BY-NC 4.0 for research and non-commercial use, with attribution, the same as the main repository. For a commercial licence, email support@loopai.com. The licence texts and the base model's attribution are in LICENSE, LICENSE-APACHE-2.0 and NOTICE.

OpenJev is an independent project, not affiliated with TypeSafe; Jev is their product.

Downloads last month: -

GGUF

Model size

9B params

Architecture

qwen35

Hardware compatibility

4-bit

5-bit

6-bit

8-bit

Model tree for openjev/OpenJev-Flash-9B-GGUF

Base model

Qwen/Qwen3.5-9B-Base

Finetuned

Qwen/Qwen3.5-9B

Finetuned

openjev/OpenJev-Flash-9B

Quantized

(4)

this model