OmniMapBench: leaderboard
Metric: Final score: overall strict exact-match accuracy on manually annotated single-choice, multiple-choice and ordering questions over map documents, chain-of-thought prompting on all 2,096 questions (%); higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 25 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 75.03 |
| 2 | Qwen 3.5 397B A17B | 72.22 |
| 3 | Qwen 3.5 Plus | 70.8 |
| 4 | Qwen 3.5 27B | 68.62 |
| 5 | Qwen 3.5 122B A10B | 68.37 |
| 6 | Kimi K2.5 | 67.64 |
| 7 | Qwen 3.5 35B A3B | 64.75 |
| 8 | Qwen 3.5 Flash | 64.56 |
| 9 | Gemini 2.5 Pro | 61.4 |
| 10 | GPT-5 | 56.54 |
| 11 | GPT-5 Mini | 56.11 |
| 12 | Gemma 4 31B (IT) | 51.67 |
| 13 | GLM-4.5V | 49.33 |
| 14 | GPT-4.1 Mini | 45.27 |
| 15 | Claude Sonnet 4.5 | 45.04 |
Interactive version: theaggregate.ai/benchmark?slug=omnimapbench · How It Works · Data refreshed daily, snapshot 2026-09-29.