M2-Verify-Gen - Statistics: leaderboard

M2-Verify (2026) is a multimodal scientific claim consistency benchmark: a model sees a figure, its caption and a claim and decides zero-shot whether the claim is supported or refuted. This board is macro-F1 in percent on the Statistics (stat) claims of the M2-Verify-Gen test set; higher is better.

Metric: Macro-F1 (%) of binary claim consistency verification on the Statistics (stat) questions of the M2-Verify-Gen test set, zero-shot; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 6 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Qwen 2.5 VL 7B Instruct65.6#1902
2InternVL3-8B60.8#1782
3Mistral Small 3.152.3#1706
4Llama 3.2 11B Vision50.3#4264
5Phi-4 Multimodal Instruct44.5#4154
6Pixtral-12B41.2#2553

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

Interactive version: theaggregate.ai/benchmark?slug=m2-verify-gen-statistics · How It Works · Data refreshed daily, snapshot 2026-11-21.