SpecVQA: 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 is the unweighted mean accuracy of the four language-and-level splits, in percent; higher is better.
Metric: Accuracy (0-1, times 100), the unweighted mean of the English and Chinese descriptive and reasoning splits; 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) | 78.72 | #156 | |
| 2 | Gemini 3 Pro (Preview) | 77.87 | #123 | |
| 3 | Gemini 2.5 Pro | 76.15 | #280 | |
| 4 | Gemini 2.5 Flash | 72.74 | #466 | |
| 5 | GPT-5 | High | 70.59 | #185 |
| 6 | O4 Mini (2025-04-16) | 70.54 | #375 | |
| 7 | O3 (2025-04-16) | 70.31 | #238 | |
| 8 | GPT-5.1 | Non-reasoning | 65.78 | #248 |
| 9 | GPT-5.2 | Non-reasoning | 65.4 | #190 |
| 10 | Claude Sonnet 4.5 | Thinking | 59.41 | #267 |
| 11 | Doubao-Seed-1.6-Flash | Thinking | 58.67 | #1268 |
| 12 | Claude Sonnet 4 | Thinking | 56.88 | #414 |
| 13 | Qwen 3 VL 8B (Thinking) | 56.31 | #924 | |
| 14 | Doubao-Seed-1.6 | Thinking | 56.02 | |
| 15 | Qwen 3 VL 8B Instruct | 51.58 | #1043 |
No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=specvqa · How It Works · Data refreshed daily, snapshot 2026-11-22.