VideoGAIA - Geography: leaderboard
Metric: Pass@1 accuracy (%; the 63 geography tasks; unified ReAct agent loop: 20 uniformly sampled frames plus web search, page visit and a thinking-with-videos tool that samples up to 20 more frames from a chosen segment, at most 40 steps; answers judged against the human-verified reference by GPT-5.5 (DeepSeek-V4-Pro fallback); mean of three runs). Source: arxiv.org. Saturation forecast: Around 2032. 20 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Seed 2.0 Pro | 60.32 |
| 2 | Qwen 3.7 Plus | 58.73 |
| 3 | Qwen 3.5 397B A17B | 52.38 |
| 4 | Gemini 3.1 Pro (Preview) | 49.21 |
| 5 | Kimi K3 | 49.21 |
| 6 | Qwen 3.5 Plus | 47.62 |
| 7 | MiMo-V2.5 | 47.62 |
| 8 | GPT-5.5 | 46.03 |
| 9 | Kimi K2.5 | 46.03 |
| 10 | Kimi K2.6 | 46.03 |
| 11 | Qwen 3.6 Plus | 46.03 |
| 12 | GLM-5V Turbo | 44.44 |
| 13 | GPT-5.4 | 42.86 |
| 14 | Qwen 3.8 Max | 42.86 |
| 15 | GLM-4.6V | 42.86 |
Interactive version: theaggregate.ai/benchmark?slug=videogaia-geography · How It Works · Data refreshed daily, snapshot 2026-09-26.