SPUR - Quantitative Reasoning: leaderboard

SPUR (2026) tests multimodal models with multiple-choice questions on multi-panel biomedical experimental figures from PubMed Central. Accuracy in percent on quantitative reasoning questions, which ask the model to check group differences, effect sizes and significance by calculation; higher is better.

Metric: Accuracy (%) on SPUR quantitative reasoning multiple-choice questions about multi-panel biomedical experimental figures from PubMed Central papers (questions drafted by GPT-4o, kept only when GPT-4o failed them in at least six of ten text-only attempts, then expert-reviewed): the questions ask the model to verify inter-group differences, effect sizes and significance by numerical comparison and calculation; higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 19 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Claude 3.7 SonnetThinking59.96#470
2Gemini 3 Pro (Preview)58.9#119
3GLM-4.5V58.48#754
4Gemini 2.5 Pro (Preview 06-05)57.94#213
5Llama 4 Maverick57.02#1110
6Ministral-3-14B-Instruct-251256.81#1639
7GPT-5.156.36#240
8O4 MiniHigh56.36#368
9Qwen 3 VL 30B A3B (Thinking)54.17#719
10Seed-1.653.89#469
11Ministral-3-8B-Instruct-251253.53#1907
12Qwen 2.5 VL 72B Instruct52.51#841
13Qwen 3 VL 30B A3B Instruct51.41#895
14Gemma 3 27B (IT)51.38#1324
15InternVL3-78B51.06#759

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

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