Note Our evaluation set: 1,420 development evaluation reports. Document labels (silver: GLM-5.3-Flash; plus pipeline labels) and 190k tagged excerpts. Used by experiments 01 and 02.

Note Decision Index: community leaderboard scoring open decision models and Jev on one suite.

Note typed-decisions benchmark: Choice/Score/Noul items with Decider 1, d1 and Jev results.

Note Fastino's 17-domain decision suite (dev split public).

deepseek-ai/DeepSeek-V4.1-Flash

Image-Text-to-Text • 763B • Updated 6 days ago • 1.16M • • 4.18k

Note LLM baseline and judge in experiment 01, via HF Inference Providers.

fastino/GLiNER2.5-Decide

Token Classification • 0.5B • Updated 9 days ago • 68.4k • 380

Note DeBERTa-v3 encoder, 340M, Apache-2.0. Native multi-label; runs on CPU.

convaiinnovations/laya

Text Classification • 0.4B • Updated 3 days ago • 20.4k • 5.28k

Note Laya: ModernBERT decision model, 421M (322M multilingual).

heman10x/rlcd-modernbert-151m

Text Classification • 0.2B • Updated 16 days ago • 29k • 38

Note openJev Verdict: ModernBERT-base, 151M; trains on Apple MPS.

MaziyarPanahi/ModernJEV-Decide-Preview

Text Classification • 0.1B • Updated 6 days ago • 158 • 13

Note ModernBERT-base, 150M, trained on 60k agent decisions.

wfzyx/von

Zero-Shot Classification • 0.4B • Updated 5 days ago • 90.7k • 57

Note Von: ModernBERT-large decision head, 395M.

hotchpotch/bekko-system-one-v0-400m

Updated 6 days ago • 7

Note bekko: small encoder decision model (17M–395M family).

BarraHome/Decision-Jef-0.1

Text Classification • 0.3B • Updated 12 days ago • 491 • 2

Note Decision-Jef: 307M encoder decision model.

jaredpalmer/kev-0.8b

Text Classification • Updated 5 days ago • 14.2k • 40

Note Kev: Qwen3.5 + LoRA + pointer head, Jev-compatible server; MLX on Mac.

jaredpalmer/kev-4b

Text Classification • Updated 5 days ago • 18.1k • 107

Note Kev-4B: Qwen3.5-4B, Jev-compatible, MLX on Mac.

jaredpalmer/kev-9b

Text Classification • Updated 5 days ago • 4.78k • 68

Note Kev-9B: Qwen3.5-9B.

jaredpalmer/kev-27b

Text Classification • 26B • Updated 5 days ago • 1.2k • 34

Note Kev-27B v2: full fine-tune of Qwen3.8-27B.

alibiserikbay/JevK5

Text Generation • 4B • Updated 12 days ago • 10.4k • 18

Note JevK5: Qwen3.5-4B with LoRA; near Jev on JevBench.

alibiserikbay/JevK5-GGUF

2B • Updated 11 days ago • 13.6k • 2

Note JevK5 GGUF for llama.cpp (Metal).

Mapika/decider-2b

Text Classification • 2B • Updated 12 days ago • 315k • 102

Note decider: Qwen3.5 Base decision models, 0.8B–35B.

Mapika/decider-4b

Text Classification • 4B • Updated 12 days ago • 24.2k • 21

Note decider-4b: ranks near Jev on JevBench.

togethercomputer/Tev1-4B-experimental

Text Generation • 5B • Updated 13 days ago • 3.76k • 30

Note Tev1: Together's Qwen3.5-4B decision model; up to ~24 options.

internlm/Intern-Decision-4B

Image-Text-to-Text • 5B • Updated 11 days ago • 2.43k • 77

Note Intern-Decision: Qwen3.5-4B decision model.

Contrastive-LM/CLM-v0.1-8B

Text Ranking • Updated 12 days ago • 3.9k • 743

Note CLM-8B: contrastive dual encoder on Qwen3-8B; cached option embeddings.

autotrust/JEV-27B

Text Classification • 27B • Updated 5 days ago • 168k • 55

Note autotrust JEV-27B: Qwen3.8-27B decision model.

StrandsAgents/strands-decider-2B-hobson-v19

Text Classification • Updated 6 days ago • 59

Note Strands Decider: 2B decision model for agent tool calls.

dwidlee/systemone-lite-0.5b

Text Generation • 0.5B • Updated 11 days ago • 1.91k

Note systemone-lite: Qwen2.5-0.5B fine-tune.

jinghao1632/bit-jev-2b-distilled

2B • Updated 7 days ago • 2.47k

Note bit-jev: 2B model distilled from Jev.

akhilaaa3/Jev-Omni

Text Classification • 12B • Updated 11 days ago • 3.91k • 376

Note Jev-Omni: Gemma 4 12B, multimodal decisions.

ollaya-dev/cygnet

Text Classification • Updated about 13 hours ago • 2

Note Cygnet: Gemma-4-12B; top of JevBench v1.5.4 (73.7).

perplexity-ai/pplx-decider-v1-27b

Text Classification • 26B • Updated 5 days ago • 1.51k • 93

Note pplx-decider: Perplexity's 27B decision model.

Cloudflare/clef

Image-Text-to-Text • 27B • Updated 5 days ago • 7.26k • 1.68k

Note Clef: Cloudflare's decision model.

prism-ml/Ternary-Bonsai-2-27B-gguf

Text Generation • 27B • Updated 11 days ago • 4.19M • 2.49k

Note Ternary Bonsai 2 27B (Qwen3.8-27B), base of Bonsai-Llama-Jev.