Jev Decision Index — Vision Leaderboard
decision-vision39 models · Updated October 2026
On the Jev Decision Index — Vision benchmark, amos-1-large ranks #1 with a score of 72.4. The full ranking, with cost per million tokens, is below.
| # | Model | Score | ECE | p50 latency | Input / 1M |
|---|---|---|---|---|---|
| 1 | amos-1-largeScira (mukaddamzaid) | 72.4 | 0.024 | — | — |
| 2 | vLLM-SR d3 (27B)vLLM Semantic Router | 71.3 | 0.016 | — | — |
| 3 | Perplexity Decider v1.1 (27B)Perplexity | 70.6 | 0.044 | — | $0.02 |
| 4 | JEV-27B-VLAutoTrust AI Lab | 69.6 | 0.018 | — | — |
| 5 | OneJev1.5-27BOmniJev | 68.9 | 0.019 | — | — |
| 6 | jev-vision-27be12ex2 | 68.7 | 0.059 | — | — |
| 7 | KnowLine-35B-A3B-Gen1PelaAI | 68.1 | 0.011 | — | — |
| 8 | vLLM-SR d3-flash (9B)vLLM Semantic Router | 66.9 | 0.024 | — | — |
| 9 | Solomon v1.1Doccy Health | 66.8 | 0.075 | — | — |
| 10 | Xor 1.2Juspay Technologies | 64.2 | 0.010 | — | — |
| 11 | RSI-Jev v6.1 VL 27B (effort auto)shgao | 64.0 | 0.037 | — | — |
| 12 | OneJev1.5-9BOmniJev | 63.4 | 0.019 | — | — |
| 13 | vLLM-SR d3-mini (4B)vLLM Semantic Router | 63.2 | 0.028 | — | — |
| 14 | Matilda Jev v1.5Maincode | 62.0 | 0.071 | — | — |
| 15 | JPT-9Bkirp | 61.7 | 0.074 | — | — |
| 16 | Cloudflare clef-omni (30B-A3B)Cloudflare | 59.5 | 0.042 | — | $0.15 |
| 17 | JPT-4Bkirp | 59.0 | 0.057 | — | — |
| 18 | Winnow-12BEldanRing | 58.8 | 0.082 | — | — |
| 19 | Surogate Rune 26B-A4B v3Surogate | 58.6 | 0.080 | — | — |
| 20 | Xor 26B-A4B v1.1Juspay Technologies | 58.0 | 0.083 | — | — |
| 21 | RSI-Jev v6.1 VL 4Bshgao | 57.4 | 0.041 | — | — |
| 22 | RSI-Jev v5.0 VL 3Bshgao | 57.3 | 0.023 | — | — |
| 23 | Aplomb 1EmpirioLabs AI | 57.3 | 0.049 | — | — |
| 24 | Intern-Decision-4BIntern Large Models | 56.5 | 0.014 | — | — |
| 25 | Jev-Omniakhilaaa3 | 56.1 | 0.085 | — | — |
| 26 | vLLM-SR d3-nano (2B)vLLM Semantic Router | 54.7 | 0.073 | — | — |
| 27 | Shisa DE-2 (repeat)Shisa.AI | 52.9 | 0.149 | — | — |
| 28 | NeoHorse-Jev-4BTokenRhythm | 51.1 | 0.022 | — | — |
| 29 | Intern-Decision-2BIntern Large Models | 47.8 | 0.032 | — | — |
| 30 | djevDavipar | 47.8 | 0.144 | — | — |
| 31 | razorback16 openjev diffusiongemma (NVFP4, vLLM)razorback16 | 47.7 | 0.174 | — | — |
| 32 | mmastrac diffusiongemma (vLLM PR 57250)mmastrac | 46.3 | 0.136 | — | — |
| 33 | OneJev1.5-0.8BOmniJev | 44.6 | 0.042 | — | — |
| 34 | vLLM-SR d3-lite (0.8B)vLLM Semantic Router | 42.1 | 0.077 | — | — |
| 35 | JPT-0.8Bkirp | 42.1 | 0.065 | — | — |
| 36 | Winnow-E4BEldanRing | 42.1 | 0.050 | — | — |
| 37 | OneJev-0.8BOmniJev | 41.5 | 0.015 | — | — |
| 38 | Intern-Decision-0.8BIntern Large Models | 36.3 | 0.061 | — | — |
| 39 | vLLM-SR d3-edge (0.6B)vLLM Semantic Router | 31.7 | 0.098 | — | — |
Frequently asked questions
What is the Jev Decision Index — Vision benchmark?
Jev Decision Index — Vision is a decision-vision used to evaluate and rank large language models. This page lists 39 models by their Jev Decision Index — Vision score.
Which model is best on Jev Decision Index — Vision?
amos-1-large (Scira (mukaddamzaid)) currently leads the Jev Decision Index — Vision leaderboard with a score of 72.4.
What is the best value model on Jev Decision Index — Vision?
Pricing data is being matched to these models.
Pricing is indicative — confirm with the provider before production use. Updated October 2026.