MicroVQA: leaderboard
Each model answers all 1,042 multiple-choice questions about biological microscopy images (one or more images each), written by biology experts and hardened against text-only shortcuts; the score is accuracy (%), from the authors' own 2025 runs. Unlike MMMU-Pro or other college-level science sets, the questions are research-level and cover image understanding, hypothesis generation and experiment proposal.
Metric: Accuracy (%). Source: jmhb0.github.io. Saturation forecast: Estimated already saturated. 14 models tracked. Updated 17 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 (2024-12-17) | 52.8 | #314 |
| 2 | pixtral-large-2411 | 49.8 | #1284 |
| 3 | Qwen 2 VL 7B Instruct | 48.8 | #2594 |
| 4 | Grok 2 (1212) | 48.4 | #1082 |
| 5 | Qwen 2 VL 72B Instruct | 47.5 | #1382 |
| 6 | VILA1.5-40B | 47.5 | #3238 |
| 7 | Claude 3.5 Haiku (20241022) | 47.1 | #1589 |
| 8 | Gemini 1.5 Flash-8B (001) | 46.7 | #2311 |
| 9 | GPT-4o Mini (2024-07-18) | 46.2 | #1494 |
| 10 | Pixtral-12B-2409 | 45.6 | #2537 |
| 11 | Llama 3.1 Nemotron 70B Instruct | 44.2 | #1471 |
| 12 | Llama 3.2 90B Vision Instruct | 42.4 | #1791 |
| 13 | VILA1.5-13B | 41.8 | #7029 |
| 14 | Llama 3.2 11B Instruct | 30.3 | #5226 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=microvqa · How It Works · Data refreshed daily, snapshot 2026-11-17.