Open decision models and reproductions with decision models

Apache-2.0 local decision model (0.8B Qwen fine-tune on llama.cpp) behind the Jev request shape, calibrated and CPU-only.

Small open vision-language decision models (2B, 4B, 9B) that answer typed questions about photos and records with a trained can't-tell.

0.8B open System One model with task adapters that makes millisecond decisions in front of a larger local model like Qwen3.8-27B.

Self-hosted drop-in replacement for Jev's System One API, powered by the 400M-parameter GLiFormer model.

Local /v1/systemone server in Japanese docs with LFM, Sarashina and fine-tuned ModernBERT-Ja backends, including image input.

AutoTrust's Apache-2.0 student of Jev 1.13 on Qwen3.8-27B, serving typed decisions and ordinary generation from one vLLM engine.

AutoTrust's vision version of JEV-27B, ranked first of 20 on the community Jev Decision Index vision board.

Local Jev-compatible /v1/systemone server answering Noul, Choice and Score from open-weight models, usable with the official SDK.

Rust engine that answers noul, choice and score from any LLM in one prefill via logprobs, wire-compatible with /v1/systemone and MCP.

Open 0.8B and 2B Qwen3.5 decision models with a local systemone-compatible server, MCP tools, agent skills and a Claude Code guard.

Makes any logprob-capable model answer typed questions Jev-style in one pass, in Jev's API shape, with probes and scorecards.

Scores every allowed option of a typed schema from one MLX prefill on Apple Silicon and assembles schema-valid JSON with probabilities.

Small open decision models (0.8B and 4B, from Qwen3.5) with a server that speaks Jev's /v1/systemone format in one forward pass.

Node.js and TypeScript package that runs Laya through ONNX Runtime with typed answers matching the Python and Jev API shapes.

Runs Laya's typed-decisions checkpoint on Mac GPUs via PyTorch MPS, with a low-memory mode down to about 0.74 GiB.

Fork of Laya that runs its choice, score and yes/no decision checkpoints on Huawei Ascend 910B NPUs as well as CPU.

Toolkit that turns Laya into a task specialist: LLM-generated and verified training data, RLCD training, calibration and longer context.

Independent MLX port of Laya for Apple Silicon: typed decisions in about 13 ms (7 ms multilingual) with no PyTorch or cloud API.

From-scratch Rust reimplementation of Laya on candle, including the encoder, decision head, calibration, RLCD training and a server.

CPU-first Rust runtime and HTTP server for Laya's multilingual model on ONNX Runtime, with portable builds and Chinese and English docs.

Pure-Rust Laya inference on candle for CPU, Metal or CUDA, with a CLI and a local web console showing close calls.

CLI and Claude Code skill that turns one typed decision into a fine-tuned, calibrated, evaluated Laya checkpoint you can serve.

Self-hosted Docker service and web console that serves Laya's multilingual model on Jev's /v1/systemone format behind API keys.

Open System One decision model (LoRA on Qwen3.5-4B) that reads typed answers from logits in one pass, with its benchmark harness.

Zero-training toolkit that turns local LLMs and multimodal models into Jev-style decision services via prefill-only scoring.

Open recipe: describe a decision task, have an LLM write training data, and train a Qwen LoRA plus decision head with calibrated outputs.

HTTP server for Mica, a 4B open decision model with 32k context serving yes/no, choice and score in TypeSafe's /v1/systemone format.

TokenRhythm's NeoHorse repo: open 4B/9B agent models plus NeoHorse-Jev-4B, a prefill-only Choice/Noul/Score decision model.

Hugging Face card for NeoHorse-Jev-4B: Apache-2.0 4B decision model for text or image-plus-text, served at /v1/systemone.

Serves the Jev request shape from any OpenAI-compatible endpoint by reading single-letter option probabilities from logprobs.