Open LMM Reasoning - MathVerse - Analytic: leaderboard
The "MathVerse - Analytic" subtask of Open LMM Reasoning. A task from the Open LMM Reasoning leaderboard, which evaluates large multimodal models on multi-step visual, mathematical and logical reasoning problems.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 100 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Doubao-1.5-Pro | 60.5 | #287 |
| 2 | GPT-4.1 Mini | 58.1 | #355 |
| 3 | InternVL3-78B | 58.1 | #352 |
| 4 | Moonshot v1 8K | 55.8 | #617 |
| 5 | GPT-4o ChatGPT | 55.8 | |
| 6 | Claude 3.7 Sonnet | 53.5 | #249 |
| 7 | GPT-4.1 (2025-04-14) | 51.2 | #250 |
| 8 | Gemini 1.5 Pro (002) | 48.8 | #349 |
| 9 | VLAA-Thinker-Qwen2.5VL-7B | 46.5 | #555 |
| 10 | VLM-R1-3B-Math-0305 | 46.5 | #743 |
| 11 | Gemini 2.0 Flash | 44.2 | #342 |
| 12 | InternVL3-38B | 44.2 | #405 |
| 13 | InternVL3-14B | 44.2 | #505 |
| 14 | Qwen 2 VL 72B | 44.2 | #460 |
| 15 | QVQ-72B-Preview | 44.2 | #440 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=open-lmm-reasoning-mathverse-analytic · How It Works · Data refreshed daily, snapshot 2026-10-20.