MMVU (Massive Multi-discipline Video Understanding) — Video-understanding multi-discipline benchmark; raws absorb 'MMVU (Pass@1)'.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Kimi K3 | Moonshot | 82.1 | 1 | 2026-08-24 |
| 2 | GPT-5.6 Sol | OpenAI | 81.2 | 1 | 2026-08-24 |
| 3 | GPT-5.2 | OpenAI | 80.8 | 1 | 2026-06-15 |
| 4 | Kimi K2.5 | Moonshot | 80.4 | 2 | 2026-08-23 |
| 5 | Claude Opus 4.8 | Anthropic | 79.2 | 1 | 2026-08-24 |
| 6 | Gemini 3 Pro | 78.9 | 2 | 2026-06-15 | |
| 7 | Claude Opus 4.5 | Anthropic | 77.3 | 1 | 2026-06-15 |
| 8 | Qwen3.5 122B A10B | Alibaba | 74.7 | 1 | 2026-08-23 |
| 9 | Qwen3.5 27B | Alibaba | 73.3 | 1 | 2026-08-23 |
| 10 | Qwen3.5 35B A3B | Alibaba | 72.3 | 1 | 2026-08-23 |
| 11 | Qwen3 VL 235B A22B Reasoning | Alibaba | 71.1 | 1 | 2026-06-15 |
| 12 | Qwen3.5 9B | Alibaba | 68.6 | 1 | 2026-06-15 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.