OpenVLM MME - OCR: leaderboard
OpenCompass OpenVLM MME subcategory score for optical character recognition tasks.
Metric: Score. Source: huggingface.co. Saturation forecast: Estimated already saturated. 234 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternVL3-8B | 200 | #620 |
| 2 | InternVL3-14B | 200 | #505 |
| 3 | Claude 3.5 Sonnet (20241022) | 192.5 | #291 |
| 4 | Gemini 2.0 Flash | 192.5 | #342 |
| 5 | InternVL3-38B | 192.5 | #405 |
| 6 | InternVL3-78B | 192.5 | #352 |
| 7 | MiniCPM-V-2.6 | 192.5 | #836 |
| 8 | Qwen 2.5 VL 7B | 192.5 | #643 |
| 9 | InternVL2.5-78B | 192.5 | #431 |
| 10 | GPT-4.1 Mini | 185 | #355 |
| 11 | GPT-4.1 (2025-04-14) | 185 | #250 |
| 12 | Gemma 3 27B | 185 | #607 |
| 13 | Qwen 2 VL 72B | 177.5 | #460 |
| 14 | Grok 2 (1212) | 177.5 | #496 |
| 15 | GPT-4.1 Nano | 170 | #731 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mme-ocr · How It Works · Data refreshed daily, snapshot 2026-10-20.