General

Artificial Analysis Decision Index leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap
License: https://creativecommons.org/licenses/by-nc/4.0/

As of 2026-10-29, the highest measured score on Artificial Analysis Decision Index is 65.2% by d1-ultra. 88 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	d1-ultra	65.2%	measured
2	mammoth-1	64.8%	estimated ± 3.7 pp, low confidence
3	elephant-1	64.7%	measured
4	Drex v2.0	59.3%	estimated ± 3.7 pp, medium confidence
5	Jev 1.14	57.1%	measured
6	Blink v0.3 26B-A4B (NVFP4)	57.1%	estimated ± 3.7 pp, medium confidence
7	Torchcast Decision 27B	57.0%	estimated ± 3.7 pp, medium confidence
8	Surogate Rune 26B-A4B v3	56.9%	estimated ± 3.7 pp, medium confidence
9	Perplexity Decider v1.2 27B	56.2%	measured
10	Drex v1.5	55.4%	estimated ± 3.7 pp, medium confidence
11	GEV-26B-Decide	54.1%	estimated ± 3.7 pp, medium confidence
12	GLiDE	53.9%	measured
13	Microsoft-Decision-1	53.8%	measured
14	Xor 1.2	53.7%	estimated ± 3.7 pp, medium confidence
15	Bespoke Nimble 9B v3	53.0%	estimated ± 3.7 pp, medium confidence
16	Eikos-27B	52.9%	estimated ± 3.7 pp, medium confidence
17	JEV-27B	52.7%	estimated ± 3.7 pp, medium confidence
18	Mercury Decide	52.7%	measured
19	Celeris-1 Decision	52.3%	measured
20	vLLM-SR Decision 2.0 Vega 27B	51.7%	estimated ± 3.7 pp, medium confidence
21	simple-jev · Qwen3.8-27B	51.4%	estimated ± 3.7 pp, medium confidence
22	JADE	51.3%	estimated ± 3.7 pp, medium confidence
23	Kev 27B	51.2%	estimated ± 3.7 pp, medium confidence
24	Quyet-1.0-Large	51.2%	estimated ± 3.7 pp, medium confidence
25	deck31b (Gemma 4 31B)	50.6%	estimated ± 3.7 pp, medium confidence
26	Decider chat · Gemma-4-31B	50.6%	estimated ± 3.7 pp, medium confidence
27	d1	50.3%	measured
28	Perplexity Decider Flash v1.2 4B	50.2%	measured
29	decider-35b-a3b	49.4%	estimated ± 3.7 pp, medium confidence
30	Decider chat · Qwen3.6-27B	49.4%	estimated ± 3.7 pp, medium confidence
31	Clef	48.6%	measured
32	JPT-35B-A3B	48.1%	estimated ± 3.7 pp, medium confidence
33	JoshuaSP diffusiongemma (open-jev)	47.6%	estimated ± 3.7 pp, medium confidence
34	decisio 0.4.0 (Qwen3.6-35B-A3B)	47.0%	estimated ± 3.7 pp, medium confidence
35	NeoHorse-Jev-4B	46.8%	estimated ± 4.5 pp, low confidence
36	Jevstral 8B	46.8%	estimated ± 4.5 pp, low confidence
37	lev	46.3%	estimated ± 3.7 pp, medium confidence
38	Clef-Flash	45.8%	measured
39	decisio 0.4.0 (Gemma 4 12B)	45.6%	estimated ± 3.7 pp, medium confidence
40	Lev1 3.7B	45.6%	estimated ± 3.7 pp, medium confidence
41	JPT-0.8B	45.1%	estimated ± 5.4 pp, low confidence
42	vLLM-SR Decision 2.0 Eos 0.8B	45.1%	estimated ± 5.4 pp, low confidence
43	ezjev 4B s2	44.8%	estimated ± 3.7 pp, medium confidence
44	Jevfire	44.6%	estimated ± 3.7 pp, low confidence
45	vLLM-SR Decision 2.0 Lux 9B	44.6%	estimated ± 3.7 pp, low confidence
46	djev	44.4%	estimated ± 3.7 pp, low confidence
47	Hopper 12B (trained)	44.3%	estimated ± 3.7 pp, low confidence
48	Clef-Omni	44.3%	measured
49	razorback16 openjev diffusiongemma (NVFP4, vLLM)	43.0%	estimated ± 3.7 pp, low confidence
50	Ryotide Qwen v0.3.1	42.9%	estimated ± 3.7 pp, low confidence
51	JevK5 9B v0.3.3	42.3%	estimated ± 3.7 pp, low confidence
52	Jobe	42.3%	estimated ± 3.7 pp, low confidence
53	Winnow-12B Q8	42.3%	estimated ± 3.7 pp, low confidence
54	JEV-9B	41.8%	estimated ± 3.7 pp, low confidence
55	Decider 4B	41.7%	estimated ± 3.7 pp, low confidence
56	Bespoke Nimble 9B v2	41.4%	estimated ± 3.7 pp, low confidence
57	Hopper (G) 1.2	41.2%	estimated ± 3.7 pp, low confidence
58	Bosun v3.1 1.7B	41.1%	estimated ± 6.6 pp, low confidence
59	JevK5 v0.3	40.7%	estimated ± 3.7 pp, low confidence
60	JPT-9B	40.5%	estimated ± 3.7 pp, low confidence
61	mmastrac diffusiongemma (vLLM PR 57250)	40.5%	estimated ± 3.7 pp, low confidence
62	SemIf	40.3%	estimated ± 3.7 pp, low confidence
63	Intern-Decision-4B	40.2%	estimated ± 3.7 pp, low confidence
64	openvons	40.0%	estimated ± 3.7 pp, low confidence
65	mini-jev	39.2%	estimated ± 3.7 pp, low confidence
66	vLLM-SR Decision 2.0 Nox 4B	39.2%	estimated ± 3.7 pp, low confidence
67	Jev-Omni	39.2%	estimated ± 3.7 pp, low confidence
68	Kas 4B	38.9%	estimated ± 3.7 pp, low confidence
69	Solomon v1.1	38.0%	estimated ± 3.7 pp, low confidence
70	JPT-4B	37.9%	estimated ± 3.7 pp, low confidence
71	Sieve 9B	37.0%	estimated ± 3.7 pp, low confidence
72	vLLM-SR Decision 2.0 Sol 2B	36.8%	estimated ± 4.9 pp, low confidence
73	Kodiak v0.2 1B (accuracy)	36.6%	estimated ± 6.6 pp, low confidence
74	Bekko System One v0 68M	35.8%	estimated ± 6.6 pp, low confidence
75	Gevva E2B	35.8%	estimated ± 4.9 pp, low confidence
76	Gevva E4B	35.5%	estimated ± 3.7 pp, low confidence
77	Bosun v3.1 0.6B	35.5%	estimated ± 6.6 pp, low confidence
78	Winnow-E4B	35.4%	estimated ± 3.7 pp, low confidence
79	EXAONE-4.0-1.2B-JEV v0.3	35.3%	estimated ± 4.9 pp, low confidence
80	vLLM-SR Decision 2.0 Kai 0.6B	34.8%	estimated ± 6.6 pp, low confidence
81	Tev1-4B-experimental	34.5%	estimated ± 3.7 pp, low confidence
82	Intelif Qwen3-4B	34.4%	estimated ± 3.7 pp, low confidence
83	Sifr 0.8B v3.1	34.3%	estimated ± 6.6 pp, low confidence
84	Jev-Style 0.8B v3	31.7%	estimated ± 6.6 pp, low confidence
85	Bekko System One v0 400M	31.6%	estimated ± 3.7 pp, low confidence
86	decider-2b	31.4%	estimated ± 3.7 pp, low confidence
87	Von 1.3	31.2%	estimated ± 6.6 pp, low confidence
88	JevK5 Lite	30.8%	estimated ± 6.6 pp, low confidence
89	Sieve 2B	30.5%	estimated ± 3.7 pp, low confidence
90	Kev 4B r10	29.9%	estimated ± 3.7 pp, low confidence
91	GLiNER2.5-Decide	29.5%	estimated ± 6.6 pp, low confidence
92	metask-jev-4b	29.0%	estimated ± 3.7 pp, low confidence
93	RSI-Jev v5.0 VL 3B	28.9%	estimated ± 3.7 pp, low confidence
94	jeff	28.8%	estimated ± 6.6 pp, low confidence
95	Lavoir	28.8%	estimated ± 6.6 pp, low confidence
96	Kev 0.8B r15	28.5%	estimated ± 6.6 pp, low confidence
97	Intern-Decision-2B	28.1%	estimated ± 3.7 pp, low confidence
98	Kev 9B v2	28.0%	estimated ± 3.7 pp, low confidence
99	Bekko System One v0 17M	27.1%	estimated ± 6.6 pp, low confidence
100	open-jev (pngwn)	26.8%	estimated ± 3.7 pp, low confidence
101	this-that 1.2	25.5%	estimated ± 3.7 pp, low confidence

Artificial Analysis Decision Index leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General