How do RSI-Jev models differ from the hosted Jev model?

They are openly published checkpoints fine-tuned from Qwen3.5 and Qwen3.8 bases to answer typed decisions, not TypeSafe's proprietary model. They implement the same request and response shape at POST /v1/systemone, so an existing Jev client works by changing its base URL.

What hardware do RSI-Jev models need?

The README says the 27B serves on a 96GB GPU, while the 4B has MLX builds for Apple Silicon (8-bit lossless and 4-bit). The effort setting chooses depth per request: low reads layer 16, medium layer 20, high all layers, and auto stops at the first confident layer.

What does the effort setting change in RSI-Jev?

It picks how many answer-head layers a request uses: low reads layer 16, medium layer 20, high all 32 layers, and auto stops at the first layer that is confident enough; the 27B uses layers 48, 56 and 64.

How do I run an RSI-Jev model locally?

Install rsi-jev with the fast and vision extras from git, then run rsi-jev serve v6.1-vl-4b --effort auto --port 8000; a Python client class, Decider, runs the model with no server.

Do RSI-Jev models generate reasoning tokens?

No. One forward pass returns a calibrated probability for every declared option, so there are no reasoning tokens to spend and every response reports which layer answered and how confident it was.

Does RSI-Jev accept images?

Yes. The VL releases take text and images, and a Python request can attach images with the images argument; the earlier 2B releases are text only.

Are the RSI-Jev Decision Index scores independent benchmarks?

They are self-reported results on the public Decision Index 0.3 set: 65.64 for the 27B with effort auto and 50.98 for the 4B, and the official score for the 27B is still pending.

CommunityJev Models & Reimplementations

RSI-Jev

Open decision models from a self-improvement loop that publishes every experiment, failures included. They answer Noul, Choice and Score questions on the same POST /v1/systemone wire format as Jev, returning a calibrated probability per option in one forward pass; the 27B scores 65.64 on the Decision Index 0.3 set.

See highlights

Open github.com

Stars: 109
Language: HTML
License: MIT
Published by: Community
Last verified: Oct 2026

Shanghua-Gao/RSI-Jev

Organization repository on GitHub

View on GitHub

Typed-decision models (noul / choice / score) trained by a self-improving loop of AI agents — checkpoints, the code that produced them, and every version that failed.

Forks: 8
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.Homepage

About this project

A recursively self-improving research loop that builds Jev -style System One decision models. 🤗 Models · Docs · Every experiment · Release notes. AI agents propose the hypotheses, register their predictions before spending GPU time, run the experiments, and retire their own champions when the evidence says to. Every release, and every experiment that failed on the way, is published with its numbers. The loop is the next version of AutoScientists.

Quoted from the project's README. Read the full README.

Tagscommunity decision-models system-one open-weights

Works forJev for Classification Jev for Scoring

Used inJev for Software & DevTools

Languages: Python
Platforms: API macOS
Capabilities: Open sourceSelf-hostedRuns locally

Highlights

8 releases in 13 days from v1.0 to v6.1-VL, with all 496 experiments written up in EXPLORE.md.
v6.1-VL 27B scored 65.64 on the Decision Index 0.3 public set with effort auto; the 4B scored 50.98.
rsi-jev serve v6.1-vl-4b picks answer layers 16, 20 or 32 per question with 23-40ms median latencies.
The server speaks Jev's API, so existing Jev clients work after pointing base_url at it.

Watch out

Code is MIT and the published weights are Apache-2.0. Scores are self-reported public-set results and the official Decision Index score for the 27B is pending; the 4B and smaller checkpoints trade quality for size, and model weights live on Hugging Face.

Questions about this entry

Short answers drawn from the linked project or page.

More Jev resources like this

Jebadiah

Open System One decision models (27B, 9B and 4B on Qwen chat bases, in bf16, GGUF and MLX) that answer Choice, Noul and Score questions with a probability for every option from one forward pass. The repo carries the trainer, data builders, evaluation harnesses and a record of every run.

Jev Models & Reimplementations#community#decision-models#open-weights

Community

getainode/jebadiahPython

Vev

Jev-compatible decision models with vision input: LoRA fine-tunes of Qwen3.5 at 4B and 9B serving Choice, Score and Noul on the /v1/systemone wire format, with screenshots and photos placed beside text in the state. Runs on your own NVIDIA GPU; weights are non-commercial.

Jev Models & Reimplementations#community#vision#open-weights

Community

Xiaooolong/vevPython

ruling

A local Jev-compatible server that reads Choice, Score and Noul answers from the logits of any MLX checkpoint or OpenAI-compatible endpoint with no training, serving POST /v1/systemone as a drop-in for the official TypeSafe SDK.

Jev Models & Reimplementations#community#mlx#calibration

Community

bradAGI/rulingPython

Rizzo Flow

877

A local server that reads next-token probabilities from a fine-tuned Spark model and returns Choice, Score, and Noul answers with zero generated tokens.

Jev Models & Reimplementations#community#python#open-models

Community

Rizzo-AI-Academy/rizzo-flowPython

AutoJev

130

An open recipe for a 27B multimodal decision model trained with full-weight SFT on 73,000 examples over one H200, serving choice, noul and score on /v1/systemone with per-checkpoint provenance and released calibration plots.

Jev Models & Reimplementations#community#calibration#qwen

Community

denis-pplx/autojevPython

Bud Decision Studio

107

Cross-platform desktop app and serving system for open Jev-like decision models. It runs laya, kev, Jev-Omni and others locally behind the Jev API at POST /v1/systemone, with a playground, versioned templates, decision history and a LoRA Train page.

Jev Models & Reimplementations#community#desktop-app#local-ai

Community

BudEcosystem/Bud-Decision-StudioHTML

From the Jev guides

Original write-ups on what this entry is for.

Multimodal Decision Models: What Clef and d1 Change

Oct 12, 2026

Jev takes text only; Clef-omni adds image, audio, and video input and Liquid d1 adds images, each keeping the same typed-answer shape. Where that matters, where it does not, and the costs the vendors underplay.

Jev Models & Reimplementations#multimodal#open-models#decision-model

Read article

The Decision-Model Price War

Oct 11, 2026

One model in September, a dozen by October. Clef, Liquid d1, Perplexity, and OpenAI all ship the same typed-answer shape now. How to read the rival benchmarks and pick on your own task.

Jev Models & Reimplementations#decision-model#benchmarks#pricing

Read article

What Is Actually on Hugging Face

Oct 8, 2026

Hundreds of repositories carry the system-one tag, but the ecosystem is a dozen ideas in different sizes. A tour of the families, the licenses, and what a download does not include.

Jev Models & Reimplementations#open-models#hugging-face#local

Read article

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai