SpecVQA - English 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 English reasoning (L1) questions. Accuracy in percent; higher is better.

Metric: Accuracy (0-1, times 100) on the English 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

#ModelSettingScoreOverall rank
1Gemini 3 Pro (Preview)77.59#119
2Gemini 3 Flash (Preview)77.59#152
3Gemini 2.5 Pro74.86#276
4O3 (2025-04-16)71#231
5Gemini 2.5 Flash70.43#463
6GPT-569.77#178
7O4 Mini (2025-04-16)68.36#370
8GPT-5.1Non-reasoning63.47#240
9GPT-5.2Non-reasoning63.28#185
10Doubao-Seed-1.6-FlashThinking56.87#1250
11Claude Sonnet 4.5Thinking55.18#261
12Doubao-Seed-1.6Thinking54.99
13Qwen 3 VL 8B (Thinking)53.48#910
14Claude Sonnet 4Thinking52.82#410
15Qwen 3 Omni 30B A3B (Thinking)48.68

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=specvqa-english-reasoning-l1 · How It Works · Data refreshed daily, snapshot 2026-11-21.