ReactBench (Reaction Diagrams) - Extraction: leaderboard
Metric: Answer accuracy (%) on the topological information extraction questions (reaction conditions, yields and labels tied to graph elements), over the 1,618 expert-annotated question-answer pairs on more than 1,300 real chemical reaction diagrams from chemistry journals and patents of ReactBench (single-line, multi-line, tree and graph topologies), multiple-choice or short open answers matched exactly against the key after template extraction, no LLM judge; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 24 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 VL 32B (Thinking) | 88.89 |
| 2 | GPT-5.5 | 88.41 |
| 3 | Qwen 3 VL 32B Instruct | 86.96 |
| 4 | Qwen 3 VL 8B (Thinking) | 85.99 |
| 5 | Qwen 2.5 VL 7B | 85.75 |
| 6 | Gemini 3.1 Pro (Preview) | 85.27 |
| 7 | Qwen 3 VL 8B Instruct | 85.02 |
| 8 | Claude 3.5 Sonnet | 81.88 |
| 9 | GPT-4o | 79.71 |
| 10 | InternVL2.5-78B | 79.47 |
| 11 | Gemini 1.5 Pro | 75.6 |
| 12 | Claude Sonnet 4.6 | 75.36 |
Interactive version: theaggregate.ai/benchmark?slug=reactbench-reaction-diagrams-extraction · How It Works · Data refreshed daily, snapshot 2026-10-07.