OMIBench - Chemistry: leaderboard
Metric: GPTScore (%, binarized LLM-judge equivalence of the final answer with the gold answer, micro-averaged) on the 217 chemistry problems of OMIBench (Olympiad-level multi-image problems from biology, chemistry, mathematics and physics competitions), zero-shot; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 37 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 VL 235B A22B (Thinking) | 33.18 |
| 2 | O4 Mini | 32.41 |
| 3 | Qwen 3 VL 30B A3B (Thinking) | 32.26 |
| 4 | GPT-5 | 29.03 |
| 5 | Gemini 3 Pro (Preview) | 25.35 |
| 6 | Qwen 3 VL 32B (Thinking) | 24.88 |
| 7 | GPT-5 Mini | 24.42 |
| 8 | Gemini 2.5 Pro | 23.96 |
| 9 | GPT-4o | 22.58 |
| 10 | Qwen 3 VL 235B A22B Instruct | 22.58 |
| 11 | GPT-4o Mini | 21.66 |
| 12 | InternVL3-78B | 20.74 |
| 13 | InternVL3-14B | 20.74 |
| 14 | Qwen 3 VL 32B Instruct | 20.74 |
| 15 | Qwen 2.5 VL 72B Instruct | 19.82 |
Interactive version: theaggregate.ai/benchmark?slug=omibench-chemistry · How It Works · Data refreshed daily, snapshot 2026-10-07.