MMBench, English v1.1 subset — Explicit version (v1.1) and metric (Acc) axes distinguish from the unversioned MMBench (EN).
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Qwen2.5 VL 72B | Alibaba | 88.3 | 1 | 2026-06-05 |
| 2 | Kimi VL A3B Thinking | Moonshot | 84.4 | 2 | 2026-06-12 |
| 3 | GPT-4o | OpenAI | 83.1 | 2 | 2026-06-05 |
| 4 | Kimi VL A3B Instruct | Moonshot | 83.1 | 2 | 2026-06-12 |
| 5 | Gemma 3 27B | 78.9 | 1 | 2026-06-05 | |
| 6 | GPT-4o mini | OpenAI | 77.1 | 1 | 2026-06-05 |
| 7 | Gemma 3 12B | 74.6 | 2 | 2026-06-05 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.