OpenVLM MMBench V1.1 CN - FP-C: leaderboard
OpenCompass OpenVLM MMBench V1.1 Chinese split accuracy for cross-instance fine-grained perception questions.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 284 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 94 | #98 |
| 2 | Gemini 2.5 Pro | 92.9 | #152 |
| 3 | GPT-5 Mini (2025-08-07) | 92.1 | #172 |
| 4 | InternVL3-78B | 89.9 | #352 |
| 5 | GPT-4.5 | 88.8 | #274 |
| 6 | GPT-4.1 (2025-04-14) | 88 | #250 |
| 7 | Qwen 2 VL 72B | 88 | #460 |
| 8 | GPT-4o ChatGPT | 88 | |
| 9 | InternVL2.5-78B | 86.5 | #431 |
| 10 | InternVL3-38B | 85.8 | #405 |
| 11 | GPT-5 Nano | 83.5 | #425 |
| 12 | Gemini 1.5 Pro (002) | 82.4 | #349 |
| 13 | Claude 3.7 Sonnet | 82.4 | #249 |
| 14 | Grok 2 (1212) | 80.9 | #496 |
| 15 | Claude 3.5 Sonnet (20241022) | 80.5 | #291 |
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-v1-1-cn-fp-c · How It Works · Data refreshed daily, snapshot 2026-10-20.