SpecVQA - Chinese Reasoning (L1): leaderboard
SpecVQA (2026) asks questions about 620 scientific spectra (NMR, IR, XRD, Raman, mass, UV-Vis and XPS) from published papers, with expert-revised descriptive and reasoning questions in English and Chinese; GPT-o4-mini judges each answer against the reference, numeric answers within 5 percent. This board covers the Chinese reasoning (L1) questions. Accuracy in percent; higher is better.
Metric: Accuracy (0-1, times 100) on the Chinese reasoning (L1) questions; expert-revised question-answer pairs on 620 spectra (NMR, IR, XRD, Raman, MS, UV-Vis, XPS) from published papers; each answer judged correct or not by GPT-o4-mini against the reference, numeric answers within a 5 percent tolerance; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 18 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 79 | #152 | |
| 2 | Gemini 3 Pro (Preview) | 77.31 | #119 | |
| 3 | Gemini 2.5 Pro | 75.71 | #276 | |
| 4 | Gemini 2.5 Flash | 71.85 | #463 | |
| 5 | GPT-5 | High | 71.47 | #178 |
| 6 | O3 (2025-04-16) | 70.43 | #231 | |
| 7 | O4 Mini (2025-04-16) | 69.96 | #370 | |
| 8 | GPT-5.1 | Non-reasoning | 62.9 | #240 |
| 9 | GPT-5.2 | Non-reasoning | 61.58 | #185 |
| 10 | Claude Sonnet 4.5 | Thinking | 57.82 | #261 |
| 11 | Doubao-Seed-1.6-Flash | Thinking | 55.74 | #1250 |
| 12 | Claude Sonnet 4 | Thinking | 55.65 | #410 |
| 13 | Qwen 3 VL 8B (Thinking) | 55.08 | #910 | |
| 14 | Doubao-Seed-1.6 | Thinking | 54.52 | |
| 15 | Qwen 3 Omni 30B A3B (Thinking) | 47.65 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=specvqa-chinese-reasoning-l1 · How It Works · Data refreshed daily, snapshot 2026-11-21.