benchgap
Agentic · tools

VITA-Bench leaderboard

As of 2026-11-21, the highest measured score on VITA-Bench is 47.9% by Qwen3.7 Max. 89 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1K-EXAONE 2.055.6%estimated ± 10.7 pp, low confidence
2Ornith-1.0-397B54.9%estimated ± 10.7 pp, low confidence
3Muse Spark Open51.7%estimated ± 9.9 pp, low confidence
4Muse Spark 1.151.3%estimated ± 9.9 pp, low confidence
5Hy4 preview49.6%estimated ± 9.9 pp, low confidence
6Gemini 3.5 Flash49.6%estimated ± 9.9 pp, low confidence
7Qwen3.7 Max47.9%measured
8Atria Dawn Preview47.5%estimated ± 6.7 pp, low confidence
9GPT-5.6 Sol47.4%estimated ± 6.7 pp, low confidence
10Kimi K3.1 Pro47.4%estimated ± 6.7 pp, low confidence
11LongCat-2.547.4%estimated ± 6.7 pp, low confidence
12Ling 3.1 Flash47.3%estimated ± 6.7 pp, low confidence
13GPT-6 Astra47.3%estimated ± 6.7 pp, low confidence
14Kimi K347.3%estimated ± 6.7 pp, low confidence
15Claude Opus 547.2%estimated ± 6.7 pp, low confidence
16Inkling 2 Small47.1%estimated ± 6.7 pp, low confidence
17GPT-5.5 Pro47.1%estimated ± 6.7 pp, low confidence
18GPT-5.4 Pro47.0%estimated ± 6.7 pp, low confidence
19GLM-5.246.9%estimated ± 9.9 pp, low confidence
20Ornith-1.0-35B46.9%estimated ± 10.7 pp, low confidence
21Step 5 Preview46.9%estimated ± 6.7 pp, low confidence
22Claude Mythos 546.8%estimated ± 6.7 pp, low confidence
23GPT-5.6 Terra46.7%estimated ± 6.7 pp, low confidence
24ERNIE 5.5 Flash Preview46.7%estimated ± 6.7 pp, low confidence
25Kimi K2.7 Code46.6%estimated ± 9.9 pp, low confidence
26Ornith-1.5-397B46.5%estimated ± 6.7 pp, low confidence
27Muse Glimmer 30B46.4%estimated ± 9.9 pp, low confidence
28Claude Sonnet 546.1%estimated ± 6.7 pp, low confidence
29Mistral Large 446.1%estimated ± 6.7 pp, low confidence
30GPT-5.546.0%estimated ± 6.7 pp, low confidence
31Claude Opus 4.846.0%estimated ± 6.7 pp, low confidence
32Claude Opus 4.645.9%estimated ± 6.7 pp, low confidence
33MiniMax M345.8%estimated ± 6.7 pp, low confidence
34DeepSeek V4 Pro45.8%estimated ± 6.7 pp, low confidence
35dots3-note Preview45.8%estimated ± 6.7 pp, low confidence
36GPT-5.6 Luna45.8%estimated ± 6.7 pp, low confidence
37Kimi K2.645.7%estimated ± 6.7 pp, low confidence
38GPT-5.445.6%estimated ± 6.7 pp, low confidence
39Qwen3.7 Plus45.6%measured
40Sage-A1.545.4%measured
41Ling 3.0 Flash Sante44.9%estimated ± 6.7 pp, low confidence
42Claude Sonnet 4.644.9%estimated ± 10.7 pp, low confidence
43Ripple-65B Preview44.8%estimated ± 9.9 pp, low confidence
44Claude Opus 4.744.6%estimated ± 6.7 pp, low confidence
45Inkling 2 Mini44.5%estimated ± 6.7 pp, low confidence
46Qwen3.6 Plus44.3%measured
47Nemotron 3.5 Super VL 120B A12B44.2%estimated ± 6.7 pp, low confidence
48Nemotron 3.5 Super VL 120B A12B NVFP444.0%estimated ± 6.7 pp, low confidence
49Inkling-Small43.9%estimated ± 6.7 pp, low confidence
50Inkling43.8%estimated ± 6.7 pp, low confidence
51Beam 1.0 501B-A23B43.7%estimated ± 6.7 pp, low confidence
52Qwen3.5 397B43.7%measured
53Beam 1.0 501B-A23B FP843.7%estimated ± 6.7 pp, low confidence
54Step 3.7 Flash43.3%estimated ± 6.7 pp, low confidence
55Beam 1.0 501B-A23B NVFP443.2%estimated ± 6.7 pp, low confidence
56DeepSeek V4 Flash 073142.1%estimated ± 6.7 pp, low confidence
57Qwen 3.6 Max (preview)41.6%estimated ± 8.2 pp, low confidence
58Ling 3.0 Flash41.5%estimated ± 6.7 pp, low confidence
59MiMo-V2.5-Pro41.0%estimated ± 10.7 pp, low confidence
60Muse Spark41.0%estimated ± 10.7 pp, low confidence
61Ornith-1.0-9B40.3%estimated ± 10.7 pp, low confidence
62Agents-A1-4B40.3%measured
63Nemotron 3.5 Bolt 74B-A9B BF1640.2%estimated ± 6.7 pp, low confidence
64LFM2.5-2.6B40.1%estimated ± 10.7 pp, low confidence
65MiMo-V2.539.6%estimated ± 10.7 pp, low confidence
66GLM-5.138.9%estimated ± 6.7 pp, low confidence
67Agents-A138.8%measured
68Solar Open 238.7%estimated ± 9.9 pp, low confidence
69Ornith-1.5-35B-A3B38.6%estimated ± 6.7 pp, low confidence
70GPT-5.4 mini38.5%estimated ± 9.9 pp, low confidence
71Qwen3.6-27B38.4%estimated ± 8.2 pp, low confidence
72Nemotron 3.5 Bolt 74B-A9B37.9%estimated ± 6.7 pp, low confidence
73GPT-5.4 nano37.7%estimated ± 9.9 pp, low confidence
74GPT-5.237.2%estimated ± 6.7 pp, low confidence
75Qwen3.6-35B-A3B35.6%measured
76Gemini 3.1 Pro35.6%estimated ± 10.7 pp, low confidence
77MiMo-V2-Pro35.6%estimated ± 10.7 pp, low confidence
78Qwen3.5-122B-A10B35.4%estimated ± 6.7 pp, low confidence
79LLaDA2.2-mini35.0%estimated ± 10.7 pp, low confidence
80GLM-5-Turbo33.9%estimated ± 10.7 pp, low confidence
81GLM-533.7%estimated ± 6.7 pp, low confidence
82LLaDA2.2-flash32.6%estimated ± 9.9 pp, low confidence
83Qwen3.5-27B32.6%estimated ± 6.7 pp, low confidence
84Qwen3.5-35B-A3B32.6%estimated ± 6.7 pp, low confidence
85Kimi K2.532.2%estimated ± 6.7 pp, low confidence
86GLM-5V-Turbo32.2%estimated ± 10.7 pp, low confidence
87Gemini 3 Flash28.6%estimated ± 10.7 pp, low confidence
88MiniMax M2.728.2%estimated ± 10.7 pp, low confidence
89Ornith-1.5-9B27.3%estimated ± 6.7 pp, low confidence
90MiMo-V2-Omni25.6%estimated ± 10.7 pp, low confidence
91LFM3-1.7B24.7%estimated ± 9.9 pp, low confidence
92Claude Opus 4.523.3%measured
93Spark-X2.6-4B23.2%estimated ± 6.7 pp, low confidence
94LongCat-Flash-Lite-Sparse21.7%measured
95Sarvam 105B18.6%estimated ± 6.7 pp, low confidence
96DeepSeek V3.218.5%measured
97Solar Pro 418.2%estimated ± 6.7 pp, low confidence
98Claude Sonnet 4.517.0%measured
99GLM-4.715.5%measured
100Nemotron 3 Ultra12.7%estimated ± 6.7 pp, low confidence
101Nemotron 3.5 Lightning 30B A3B NVFP46.8%estimated ± 6.7 pp, low confidence
102Sarvam 30B6.1%estimated ± 6.7 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General