OpenVLM LLaVA-Bench - Detail: leaderboard
Metric: Score. Source: huggingface.co. 234 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 (2025-04-14) | 103.7 |
| 2 | Gemma 3 27B | 102.9 |
| 3 | Gemma 3 4B | 96.6 |
| 4 | GPT-4.1 Mini | 95.5 |
| 5 | Claude 3.7 Sonnet | 94.2 |
| 6 | Gemma 3 12B | 93.5 |
| 7 | Grok 2 (1212) | 93.1 |
| 8 | Gemini 2.0 Flash | 93 |
| 9 | Gemini 1.5 Pro | 91 |
| 10 | InternVL3-78B | 90.2 |
| 11 | InternVL3-38B | 83.5 |
| 12 | Claude 3.5 Sonnet (20241022) | 82.8 |
| 13 | GPT-4.1 Nano | 81.1 |
| 14 | Claude 3 Sonnet | 79.4 |
| 15 | Claude 3.5 Sonnet | 78.7 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-llava-bench-detail · How It Works · Data refreshed daily, snapshot 2026-09-05.