DecisionBench Leaderboard
decision42 models · Updated October 2026
On the DecisionBench benchmark, Bosun v3.1 1.7B ranks #1 with a score of 0.8. The full ranking, with cost per million tokens, is below.
| # | Model | Score | ECE | p50 latency | Input / 1M |
|---|---|---|---|---|---|
| 1 | Bosun v3.1 1.7BHanno Labs | 0.8 | 0.050 | — | — |
| 2 | Bosun v3.1 0.6BHanno Labs | 0.8 | 0.066 | — | — |
| 3 | imajev-4bmohit67890 | 0.8 | 0.069 | — | — |
| 4 | Winnow-12BEldanRing | 0.8 | 0.142 | — | — |
| 5 | openjev-sglang (Qwen3.6-35B-A3B)Eric Zhang | 0.7 | 0.109 | — | — |
| 6 | JevTypeSafe AI | 0.7 | 0.128 | — | $0.04 |
| 7 | razorback16 openjev diffusiongemma (NVFP4, vLLM)razorback16 | 0.7 | 0.228 | — | — |
| 8 | Open-Jev-9BZefanCai | 0.7 | 0.196 | — | — |
| 9 | djevDavipar | 0.7 | 0.229 | — | — |
| 10 | Kev 4B r10Jared Palmer | 0.7 | 0.255 | — | $0.04 |
| 11 | djev-sparkmmastrac | 0.6 | 0.193 | — | — |
| 12 | reflex 4Bkshetrajna12 | 0.6 | 0.117 | — | — |
| 13 | kev-8bJared Palmer | 0.6 | 0.229 | — | — |
| 14 | ZTC-Judge-9BFINAL_Bench | 0.6 | 0.364 | — | — |
| 15 | Decision-1.0-Eos-0.8BvLLM Semantic Router | 0.6 | 0.050 | — | — |
| 16 | Decider 2BMapika | 0.6 | 0.111 | — | — |
| 17 | Open-Jev-2BZefanCai | 0.6 | 0.211 | — | — |
| 18 | Xor 1.2Juspay Technologies | 0.6 | 0.041 | — | — |
| 19 | Tev1-4B-experimentalTogether | 0.6 | 0.111 | — | $0.04 |
| 20 | SemIfTheoLeeCJ | 0.6 | 0.129 | — | — |
| 21 | reflex Qwen3.8-27B-FP8 (pmscr)kshetrajna12 | 0.6 | 0.070 | — | — |
| 22 | open-alternative-jevikermoel | 0.6 | 0.144 | — | — |
| 23 | mojevMoLeMo-Lab | 0.6 | 0.088 | — | — |
| 24 | Decider 35B-A3BMapika | 0.6 | 0.044 | — | — |
| 25 | LitJevzhengxuyu | 0.6 | 0.083 | — | — |
| 26 | Bespoke-Nimble-9BBespoke Labs | 0.5 | 0.179 | — | — |
| 27 | cua-s1-4b-0.2cua-ai | 0.5 | 0.114 | — | — |
| 28 | jqvOctalab-Inc | 0.5 | 0.094 | — | — |
| 29 | smalljev semantic-v9isHeSatoshi | 0.5 | 0.147 | — | — |
| 30 | OpenDecisiondeepanwadhwa | 0.5 | 0.183 | — | — |
| 31 | Kev 0.6BJared Palmer | 0.5 | 0.204 | — | — |
| 32 | simple-jev · Qwen3.8-27B (featherless)featherless-ai | 0.5 | 0.218 | — | — |
| 33 | SimpleJev Qwen3.6-35B-A3B (Featherless)Featherless AI | 0.5 | 0.314 | — | $0.00 |
| 34 | kev-0.5bJared Palmer | 0.4 | 0.160 | — | — |
| 35 | ZTC-Judge-4BFINAL_Bench | 0.4 | 0.180 | — | — |
| 36 | LAYA typed-decisionsConvai Innovations | 0.4 | 0.033 | — | — |
| 37 | LAYAConvai Innovations | 0.4 | 0.142 | — | $0.00 |
| 38 | open-jev DeBERTa-v3-largecom-kotobalabs | 0.4 | 0.032 | — | — |
| 39 | NanoJevC-Tianyu | 0.4 | 0.155 | — | — |
| 40 | GLiNER2.5-DecideFastino | 0.4 | 0.169 | — | — |
| 41 | RLCD ModernBERT 151Mheman10x | 0.3 | 0.067 | — | — |
| 42 | certo-decision-modelaltslate | 0.2 | 0.041 | — | — |
Frequently asked questions
What is the DecisionBench benchmark?
DecisionBench is a decision used to evaluate and rank large language models. This page lists 42 models by their DecisionBench score.
Which model is best on DecisionBench?
Bosun v3.1 1.7B (Hanno Labs) currently leads the DecisionBench leaderboard with a score of 0.8.
What is the best value model on DecisionBench?
Pricing data is being matched to these models.
Pricing is indicative — confirm with the provider before production use. Updated October 2026.