OpenVLM MMBench V1.0 CN - LR: leaderboard
MMBench v1.0 Chinese test questions on logical reasoning (LR: structured image-text understanding and future prediction), from the OpenCompass OpenVLM Leaderboard (VLMEvalKit). CircularEval counts a question only if every rotation of its options is answered correctly. Accuracy in percent, higher is better; this is the original question set, separate from the MMBench V1.1 boards.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 230 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4.1 (2025-04-14) | 89 | #257 |
| 2 | GPT-4.1 Mini | 86.7 | #363 |
| 3 | InternVL3-38B | 83.2 | #413 |
| 4 | Claude 3.5 Sonnet (20241022) | 82.7 | #302 |
| 5 | Grok 2 (1212) | 82.1 | #502 |
| 6 | InternVL3-78B | 81.5 | #361 |
| 7 | InternVL2.5-78B | 80.9 | #434 |
| 8 | Qwen 2 VL 72B | 78 | #450 |
| 9 | InternVL3-14B | 76.9 | #506 |
| 10 | Claude 3.7 Sonnet | 75.7 | #259 |
| 11 | Moonshot v1 8K | 75.7 | #622 |
| 12 | Claude 3.5 Sonnet | 74 | #362 |
| 13 | Gemma 3 27B | 72.8 | #606 |
| 14 | InternVL3-8B | 69.4 | #624 |
| 15 | Qwen 2.5 VL 7B | 69.4 | #641 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmbench-v1-0-cn-lr · How It Works · Data refreshed daily, snapshot 2026-11-05.