CommunityJev Models & Reimplementations
NeoHorse-Jev-4B
TokenRhythm's 4B prefill-only decision model built on NeoHorse-1-4B: given a state and application-defined questions it returns Choice, Noul, and Score probabilities without generating text, reporting a 77.70 six-group aggregate under Apache-2.0.
- Published by
- Community
- Author
- TokenRhythm
- Last verified
- Oct 2026
- Languages
- Python
- Capabilities
- Open sourceSelf-hostedRuns locally
Highlights
- 4B structured decision model built on NeoHorse-1-4B, predicting Choice, Noul, and Score decisions in one prefill-only forward pass.
- Six-group text aggregate 77.70, the highest among the four open-weight decision models with complete results in the card's comparison.
- 83.26% mean accuracy across Nimble, VitaminC, and MASSIVE, 11.50 percentage points above the NeoHorse-1-4B baseline.
- Leads the comparison on Kev (81.92) and Nimble (87.23); Laya wins VitaminC and Open-Jev-9B wins JevBench and OpenJev text.
- Runs with vLLM, SGLang, or a native Python, CLI, and HTTP runtime; optional image requests pair one image with text.
Watch out
Apache-2.0, built on TokenRhythm's NeoHorse-1-4B base. The 77.70 aggregate is the card's own equal-weight mean of six group scores, not pooled per-example accuracy or an official leaderboard, and it ranks only models with all six groups; Kev and Nimble wins sit beside losses on JevBench, OpenJev text, and VitaminC.
Questions about this entry
Short answers drawn from the linked project or page.
A 4B structured decision model from TokenRhythm, built on NeoHorse-1-4B. Given a state and questions defined by your application, it predicts decisions and their probabilities for routing requests, selecting tools, checking conditions, and rating outcomes.
Three: Choice returns a selected option plus the candidate distribution for routing or tool selection; Noul returns P(true) for condition checks and workflow gates; Score returns a distribution over ordered levels with an expected value.
The card lists vLLM, SGLang, and a native Python, CLI, and HTTP runtime. The repo ships three backbone safetensors shards, a preprocessor config, and a neohorse_decision wheel, with DEPLOYMENT.md covering the setup, all under Apache-2.0.
77.70, the card's equal-weight mean of JevBench, Kev, OpenJev text, Nimble, VitaminC, and MASSIVE and the highest among the four open-weight models with complete results. Kev is 81.92 and Nimble 87.23, both best in the card's table.
On the three shared benchmarks, the card reports 83.26% mean accuracy across Nimble, VitaminC, and MASSIVE, 11.50 percentage points above the NeoHorse-1-4B baseline, whose grouped scores are 69.15, 63.27, and 82.86.
Not fully. Decisions are text-first, and optional image requests combine a single image with text. The related GGUF repo ships the visual weights and decision head inside each quantized file, and the demos cover Tetris, Snake, Mahjong, robot manipulation, a bomb arena, and driving.
It is the card's own equal-weight mean of six displayed group scores, not pooled per-example accuracy or an official combined leaderboard. Only models with all six groups are ranked by it, images and games do not enter, and protocols differ per group.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
LinkedIn: What multimodal decision models actually change, and what they do not.
LinkedInJevDirectory.org
Hacker News: Decision Index 0.2.1 scores open reproductions of Jev on a frozen suite.
jackbravo4 pointsDecision Index 0.2.1 – 70 open reproductions of TypeSafe Jev's Decision ModelRead the thread on Hacker NewsHacker News: Pointer to a DiffusionGemma plus vLLM tutorial for a Jev-shaped classifier.
1 pointHow to build a Jev-style classifier with DiffusionGemma and vLLMRead the thread on Hacker NewsHacker News: A one-token logprob wrapper that also classifies webcam frames.
allanrbo151 pointsA single function Jev-like wrapper for LLMs, including vision modelsRead the thread on Hacker NewsMore Jev resources like this
The GGUF edition of TokenRhythm's NeoHorse-Jev-4B: BF16 through Q2_K quantizations that each bundle the language backbone, visual weights, decision head, and tokenizer, plus a shared runtime for text and single-image requests.
Jev Models & Reimplementations#community#open-models#local
Community
A local System One-compatible decision model on Qwen2.5-0.5B-Instruct: 50.7% accuracy and 0.245 ECE on JevBench public with a 13.1 ms median, built for demos rather than production.
Jev Models & Reimplementations#community#open-models#local
Community
A decision model that scores its own clarifying questions by value of information: at AUC .799 against an oracle .797, it asks only when asking is expected to change the answer, then decides or hands off.
Jev Models & Reimplementations#community#open-models#local
Community
akhilaaa3's merged multimodal decision classifier on Gemma 4 12B IT: probabilities for text, image, audio or video options, 87.57% on DecisionBench Medium and MMAU 63.10%, Apache-2.0.
Jev Models & Reimplementations#community#open-models#local
Community
Ruiruiz30's Apple Silicon MLX conversion of Jev-Omni: 4-bit affine language weights on a Mac mini M4 16GB, about 7 GB peak Metal memory and roughly 963 ms text decisions, Apache-2.0.
Jev Models & Reimplementations#community#open-models#local
Community
The v3 Jev-Style decision model at 2B in GGUF: F16 3.78 GB and Q8_0 2.01 GB match PyTorch FP32 on 100% and 99.7% of parity rows, 73.6% on JevBench v1.4.1 public items.
Jev Models & Reimplementations#community#open-models#local
Community
From the Jev guides
Original write-ups on what this entry is for.
Training recipes, looped architectures, and language ports: what the open side of the typed-decision class shipped this month, and how to pick by runtime and calibration instead of the leaderboard.
Jev Models & Reimplementations#open-models#training#local
Read article
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
Taking a decision checkpoint out of Python: what the GGUF and browser WebGPU exports preserve, what they change, and the parity evidence the model cards publish.
Jev Tools & CLIs#open-models#local#onnx
Read article
From the Jev community
Posts from builders shipping with Jev right now.