openjev/OpenJev-Flash-9B-MLX

OpenJev Flash 9B, MLX 8-bit (Apple silicon, text only)

This is the language model of OpenJev Flash 9B, converted for Apple silicon with mlx-lm (8-bit, group size 64). It is about 9.5 GB. The main card covers the model, the API and all results. This page says what this build is, how it was checked and how to run it.

Text only. The converter keeps the language model and drops the vision tower, so this build answers questions about text, JSON and DOM, not screenshots.

Results

This is the build behind OpenJev Flash 9B's JevBench score. On JevBench's 231 public items it is on par with Cloudflare's Clef-Flash and ahead of Kev-9B and Nimble 9B:

model	correct	accuracy
Clef-Flash (Cloudflare)	190 of 231	82.3%
OpenJev Flash 9B (this MLX build)	188 of 231	81.4%
Kev-9B	183 of 231	79.2%
Nimble 9B (Bespoke Labs)	183 of 231	79.2%

About a quarter of a second per decision on an Apple silicon Mac. No JevBench item was used to train or tune it. More results, by family and on 10,000 text questions, are on the main card.

Run it

uv venv mlx --python 3.12
uv pip install --python mlx/bin/python "mlx==0.32.2" "mlx-lm==0.31.3" \
  "transformers==5.17.0" "openai==3.16.2" "httpx==0.28.1" "huggingface-hub==1.32.0"

# Download the text-only weights and the two helper files.
mlx/bin/hf download openjev/OpenJev-Flash-9B-MLX --local-dir OpenJev-Flash-9B-MLX
mlx/bin/hf download openjev/OpenJev-Flash-9B helper/shim.py helper/shim_mlx.py \
  --local-dir openjev-flash-9b-api

TOKENIZER=OpenJev-Flash-9B-MLX SHIM_MODEL=OpenJev-Flash-9B-MLX \
READOUT_T=1.07 READOUT_NOUL_T=1.074766 READOUT_NOUL_BIAS=0 \
READOUT_TARGETED=1 READOUT_INSTR_STYLE=pyrepr SHIM_STAGGER=1 \
  mlx/bin/python openjev-flash-9b-api/helper/shim_mlx.py \
  --helper openjev-flash-9b-api/helper/shim.py --model OpenJev-Flash-9B-MLX --port 3000

Then call http://localhost:3000/v1/systemone exactly as the main card describes. shim_mlx.py replaces only the helper's model client with an MLX one; prompts, option layout, readout and calibration are the helper's own code. The helper's built-in defaults belong to OpenJev 27B, so pass the five READOUT_* settings above. Leave --prefix-cache off to reproduce the measured path. Append --selfcheck for a quick smoke test that prints the helper hash (it should begin 81a22f1b), the calibration and a few answers, then exits.

Licence

Weights: CC BY-NC 4.0 (research and non-commercial use), the same as the main repository. For a commercial licence, email support@loopai.com. Helper and serving files: Apache 2.0. The base model, Qwen/Qwen3.5-9B, is Apache 2.0. The licence texts and the base model's attribution are in the main repository's LICENSE, LICENSE-APACHE-2.0 and NOTICE.

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