benchgap
General

Artificial Analysis Decision Index leaderboard

As of 2026-11-18, the highest measured score on Artificial Analysis Decision Index is 74.8% by d1.2-ultra. 89 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1d1.2-ultra74.8%measured
2Drex Max v1.172.6%measured
3GLiDE v1.272.5%measured
4mammoth-171.4%measured
5d1-ultra70.6%measured
6Perplexity Decider v1.3 27B70.4%measured
7Perplexity Decider v1.2 27B69.8%measured
8Drex Max69.1%estimated ± 1.3 pp, high confidence
9d1.268.7%measured
10SubQ Snap 1.167.9%measured
11deck31b (Gemma 4 31B)67.3%estimated ± 1.3 pp, high confidence
12Quyet-1.0-Large67.1%estimated ± 1.3 pp, high confidence
13Decider chat · Gemma-4-31B66.8%estimated ± 1.3 pp, high confidence
14elephant-166.4%measured
15Jev 1.1466.2%measured
16Gemini Decision 166.0%measured
17Torchcast Decision 27B65.2%estimated ± 1.3 pp, high confidence
18Kev 27B64.4%estimated ± 1.3 pp, high confidence
19Surogate Rune 26B-A4B v364.3%estimated ± 1.3 pp, high confidence
20Perplexity Decider Flash v1.3 4B64.1%measured
21Celeris-1 Decision64.0%estimated ± 2.9 pp, high confidence
22GLiDE64.0%estimated ± 2.9 pp, high confidence
23Eikos-27B63.9%estimated ± 1.3 pp, high confidence
24JEV-27B63.8%estimated ± 1.3 pp, high confidence
25Blink v0.3 26B-A4B (NVFP4)63.8%estimated ± 1.3 pp, high confidence
26vLLM-SR Decision 2.0 Vega 27B63.7%estimated ± 1.3 pp, high confidence
27JADE62.6%estimated ± 1.3 pp, high confidence
28Decider chat · Qwen3.6-27B62.5%estimated ± 1.3 pp, high confidence
29GEV-26B-Decide62.3%estimated ± 1.3 pp, high confidence
30simple-jev · Qwen3.8-27B62.2%estimated ± 1.3 pp, high confidence
31Mercury Decide62.1%measured
32Microsoft-Decision-162.1%measured
33Celeris-1.1 Decision62.0%measured
34JPT-35B-A3B61.7%estimated ± 1.3 pp, high confidence
35Drex v2.061.6%measured
36Xor 1.261.5%estimated ± 1.3 pp, high confidence
37d160.8%measured
38Clef60.3%measured
39GPT-6 Luna (Decisions)60.2%measured
40Perplexity Decider Flash v1.2 4B60.1%measured
41Bespoke Nimble 9B v359.6%estimated ± 1.3 pp, high confidence
42decisio 0.4.0 (Qwen3.6-35B-A3B)58.9%estimated ± 1.3 pp, medium confidence
43JoshuaSP diffusiongemma (open-jev)58.8%estimated ± 1.3 pp, high confidence
44Clef-Omni57.8%measured
45decisio 0.4.0 (Gemma 4 12B)57.6%estimated ± 1.3 pp, medium confidence
46decider-35b-a3b57.4%estimated ± 1.3 pp, high confidence
47GLiNER2.5-Decide57.4%estimated ± 2.6 pp, medium confidence
48Bosun v3.1 1.7B57.4%estimated ± 2.6 pp, medium confidence
49Laya57.4%estimated ± 2.6 pp, medium confidence
50Winnow-12B Q857.1%estimated ± 1.3 pp, medium confidence
51Drex v1.556.7%estimated ± 1.3 pp, medium confidence
52JevK5 9B v0.3.356.7%estimated ± 1.3 pp, medium confidence
53Clef-Flash56.6%estimated ± 1.3 pp, medium confidence
54vLLM-SR Decision 2.0 Lux 9B56.5%estimated ± 1.3 pp, medium confidence
55Hopper 12B (trained)55.5%estimated ± 1.3 pp, medium confidence
56JPT-9B55.2%estimated ± 1.3 pp, medium confidence
57ezjev 4B s254.9%estimated ± 1.3 pp, medium confidence
58JEV-9B54.5%estimated ± 1.3 pp, medium confidence
59Jevfire53.6%estimated ± 1.3 pp, medium confidence
60Kev 9B v253.5%estimated ± 1.3 pp, medium confidence
61vLLM-SR Decision 2.0 Nox 4B53.4%estimated ± 1.3 pp, medium confidence
62Bespoke Nimble 9B v253.3%estimated ± 1.3 pp, medium confidence
63Jev-Omni52.9%estimated ± 1.3 pp, medium confidence
64Sieve 9B52.6%estimated ± 1.3 pp, medium confidence
65Hopper (G) 1.252.3%estimated ± 1.3 pp, medium confidence
66JPT-4B52.3%estimated ± 1.3 pp, medium confidence
67Solomon v1.151.8%estimated ± 1.3 pp, medium confidence
68Decider 4B50.3%estimated ± 1.3 pp, medium confidence
69JevK5 v0.350.2%estimated ± 1.3 pp, medium confidence
70Intern-Decision-4B50.1%estimated ± 1.3 pp, medium confidence
71djev49.6%estimated ± 1.3 pp, medium confidence
72NeoHorse-Jev-4B49.5%estimated ± 1.3 pp, medium confidence
73razorback16 openjev diffusiongemma (NVFP4, vLLM)49.5%estimated ± 1.3 pp, medium confidence
74Kas 4B49.0%estimated ± 1.3 pp, medium confidence
75Kev 4B r1048.2%estimated ± 1.3 pp, medium confidence
76mmastrac diffusiongemma (vLLM PR 57250)47.8%estimated ± 1.3 pp, medium confidence
77lev47.4%estimated ± 1.3 pp, medium confidence
78Tev1-4B-experimental47.3%estimated ± 1.3 pp, medium confidence
79Lev1 3.7B47.0%estimated ± 1.3 pp, medium confidence
80open-jev (pngwn)46.3%estimated ± 1.3 pp, medium confidence
81Jevstral 8B46.3%estimated ± 1.3 pp, medium confidence
82Bekko System One v0 68M46.0%estimated ± 2.7 pp, medium confidence
83metask-jev-4b45.8%estimated ± 1.3 pp, medium confidence
84Winnow-E4B45.4%estimated ± 1.3 pp, medium confidence
85Intelif Qwen3-4B44.7%estimated ± 1.3 pp, medium confidence
86Ryotide Qwen v0.3.144.1%estimated ± 1.3 pp, medium confidence
87Qwen-2.5-1B-RLCD43.9%estimated ± 2.7 pp, medium confidence
88openvons42.9%estimated ± 1.3 pp, medium confidence
89Jobe42.1%estimated ± 1.3 pp, medium confidence
90Gevva E4B41.6%estimated ± 1.3 pp, medium confidence
91SemIf39.1%estimated ± 1.3 pp, medium confidence
92mini-jev37.9%estimated ± 1.3 pp, medium confidence
93this-that 1.237.3%estimated ± 1.3 pp, medium confidence
94decider-2b37.0%estimated ± 1.3 pp, medium confidence
95Gevva E2B36.6%estimated ± 1.3 pp, medium confidence
96vLLM-SR Decision 2.0 Sol 2B35.6%estimated ± 1.3 pp, medium confidence
97EXAONE-4.0-1.2B-JEV v0.334.6%estimated ± 1.3 pp, medium confidence
98Sieve 2B33.9%estimated ± 1.3 pp, medium confidence
99Intern-Decision-2B31.2%estimated ± 1.3 pp, medium confidence
100Kodiak v0.2 1B (accuracy)28.1%estimated ± 1.9 pp, medium confidence
101vLLM-SR Decision 2.0 Eos 0.8B27.8%estimated ± 1.3 pp, medium confidence
102Bekko System One v0 400M27.6%estimated ± 1.3 pp, medium confidence
103Sifr 0.8B v3.124.9%estimated ± 1.9 pp, medium confidence
104Kev 0.8B r1523.4%estimated ± 1.9 pp, medium confidence
105JPT-0.8B21.4%estimated ± 1.9 pp, medium confidence
106LFM2.5-2.6B-RLCD21.4%estimated ± 1.9 pp, medium confidence
107Intern-Decision-0.8B21.1%estimated ± 1.9 pp, medium confidence
108Von 1.320.3%estimated ± 1.9 pp, medium confidence
109vLLM-SR Decision 2.0 Kai 0.6B19.8%estimated ± 1.9 pp, medium confidence
110JevK5 Lite16.2%estimated ± 1.9 pp, medium confidence
111Tev1-0.8B-experimental15.3%estimated ± 1.9 pp, medium confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General