VITA-Bench

VITA-Bench - agentic benchmark. Avg subdomain scores; using claude-4-sonnet as judger.

Category

General Agents

Max Score

100.0

Score Type

percent

Active

Yes

Model Scores

Scores for VITA-Bench

Model Score Date Verified
Qwen3.6-35B-A3B
80.83%
16.04.2026 Verified
Gemma4-31B-it
100.00%
16.04.2026 Unverified
Qwen3.5-35B-A3B
63.99%
16.04.2026 Unverified
Gemma4-26B-A4B
84.20%
16.04.2026 Unverified
Qwen3.5-27B
96.89%
16.04.2026 Unverified
Qwen3.5-4B
45.60%
01.02.2026 Verified
Qwen3.5-122B-A10B
75.65%
01.02.2026 Verified
Qwen3-235B-A22B
70.47%
01.02.2026 Unverified
GPT-5-mini 2025-08-07
24.61%
01.02.2026 Unverified
Qwen3.5-9B
29.02%
— Verified
Spark-X2.5-4B
53.89%
— Verified
Spark-X2.5-1.7B
10.10%
— Verified
Gemma4-12B
20.73%
— Verified
Gemma4-E2B
4.40
— Verified
Qwen3.5-2B
2.07%
— Verified
Gemma4-E4B
1.04%
— Verified