LLM Stats (VITA-Bench): leaderboard
Video, Image, Text, and Audio multimodal benchmark testing understanding across all four modalities in integrated scenarios.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around June 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 397B A17B | 49.7 | #146 |
| 2 | Qwen 3.7 Max | 47.9 | #75 |
| 3 | Qwen 3.7 Plus | 45.6 | #92 |
| 4 | Qwen 3.6 Plus | 44.3 | #120 |
| 5 | Qwen 3.5 27B | 41.9 | #187 |
| 6 | Qwen 3 Max (Thinking) | 40.9 | #208 (Qwen 3 Max) |
| 7 | Qwen 3.6 35B A3B | 35.6 | #226 |
| 8 | Qwen 3.5 122B A10B | 33.6 | #173 |
| 9 | Qwen 3.5 35B A3B | 31.9 | #256 |
| 10 | Qwen 3.5 9B | 29.8 | #379 |
| 11 | Qwen 3.5 4B | 22 | #475 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-vita-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.