T2S-Bench - Functional Mapping: leaderboard
Metric: Exact match (%) on the 69 Functional Mapping questions on T2S-Bench-MR (500 expert-checked multiple-choice questions over text-structure pairs from scientific papers in six domains; single- and multiple-answer items, only the text and the question given), deterministic decoding at temperature 0; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 45 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 89.86 | #145 |
| 2 | Claude Sonnet 4 (20250514) | 81.16 | #211 |
| 3 | Claude Sonnet 4.5 | 79.71 | #138 |
| 4 | GPT-5.1 | 78.26 | #131 |
| 5 | Qwen 3 235B A22B 2507 Instruct | 78.26 | #291 |
| 6 | Gemini 2.5 Flash | 76.81 | #237 |
| 7 | Gemini 2.0 Flash | 76.81 | #331 |
| 8 | DeepSeek V3.1 | 73.91 | #260 |
| 9 | DeepSeek V3 (0324) | 73.91 | #332 |
| 10 | DeepSeek V3.2 | 72.46 | #198 |
| 11 | Claude Haiku 4.5 (20251001) | 72.46 | #251 |
| 12 | Qwen 3 235B A22B 2507 (Thinking) | 72.46 | #253 (Qwen 3 235B A22B 2507) |
| 13 | DeepSeek R1 0528 | 72.46 | #217 |
| 14 | Kimi K2 0905 | 72.46 | #282 |
| 15 | GPT-5.2 | 69.57 | #105 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=t2s-bench-functional-mapping · How It Works · Data refreshed daily, snapshot 2026-10-11.