GeoBrowse - Level 1 Country (Direct): leaderboard
Metric: Accuracy (%) on the country-level targets of GeoBrowse Level 1 (199 street-scene instances whose location must be composed from weak, fragmented visual cues), judged correct by an LLM judge with the Humanity's Last Exam judging prompt, mean of 3 runs, under direct inference: the model answers in a single pass without tool calls; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro | 53.3 |
| 2 | Claude Opus 4.5 | 43.4 |
| 3 | Gemini 2.5 Pro | 41.3 |
| 4 | Gemini 2.5 Flash | 40.2 |
| 5 | GPT-5 | 38 |
| 6 | GPT-4o | 35.7 |
| 7 | GPT-5.2 | 30.4 |
| 8 | GPT-4.1 | 30.4 |
| 9 | Llama 3.2 90B | 19.6 |
| 10 | Qwen 2.5 VL 7B | 11.1 |
Interactive version: theaggregate.ai/benchmark?slug=geobrowse-level-1-country-direct · How It Works · Data refreshed daily, snapshot 2026-10-07.