DocVQA — Results
| Rank | Model | Score |
|---|---|---|
| 1 | lfm3-vl-2b | 96.7 |
| 2 | deepseek-v4-1-flash | 95.6 |
| 3 | north-micro-vision-instruct | 92.1 |
| 4 | lfm2-5-vl-3b | 91.1 |
| 5 | claude-3-5-sonnet | 0.952 |
| 6 | qwen2-5-omni-7b | 0.952 |
| 7 | qwen2-5-vl-32b-instruct | 0.948 |
| 8 | llama-4-maverick | 0.944 |
| 9 | llama-4-scout | 0.944 |
| 10 | grok-2 | 0.936 |
| 11 | nova-pro | 0.935 |
| 12 | gpt-4o | 0.928 |
| 13 | nova-lite | 0.924 |
| 14 | llama-3-2-90b-instruct | 0.901 |
| 15 | gemma-3-12b | 0.871 |
| 16 | gemma-3-27b | 0.866 |
| 17 | grok-1-5 | 0.856 |
| 18 | gemma-3-4b | 0.758 |
D
DocVQA — Leaderboard
1 phaseActive
DocVQA Leaderboard
No self-reported entries.