GeoBench evaluates image geolocation in GeoGuessr. This DataLearner entry fixes the benchmark identity to country accuracy on the 100-image A Community World set, keeping it separate from other maps and distance-based metrics.
Browse the latest scores, model modes, release dates, and parameter sizes for GeoBench ACW.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
| Rank | Model | License | |||
|---|---|---|---|---|---|
![]() Claude Fable 5 Standard Mode | 89.00 | 2026-06-09 | Unknown | Closed | |
![]() Gemini 3.0 Flash Standard Mode | 88.00 | 2025-12-17 | Unknown | Closed | |
![]() Gemini-2.5-Pro-Preview-05-06 Standard Mode | 86.00 | 2025-05-06 | Unknown | Closed | |
4 | ![]() Gemini 3 Pro Standard Mode | 84.00 | 2026-02-01 | Unknown | Closed |
5 | ![]() Gemini 2.5 Pro Experimental 03-25 Standard ModeToolsInternet | 81.00 | 2025-03-25 | Unknown | Closed |
6 | ![]() Gemini-2.5-Pro-Preview-05-06 Standard ModeToolsInternet | 81.00 | 2025-05-06 | Unknown | Closed |
7 | ![]() GPT-5 Thinking Level · Medium | 81.00 | 2025-08-07 | Unknown | Closed |
8 | ![]() OpenAI o1 Thinking Level · Medium | 80.00 | 2024-12-05 | Unknown | Closed |
9 | ![]() Gemini 2.5 Flash Standard Mode | 76.00 | 2025-04-17 | Unknown | Closed |
10 | ![]() Opus 4.5 Standard Mode | 75.00 | 2025-11-25 | Unknown | Closed |
11 | ![]() OpenAI o3 Thinking Level · Medium | 74.00 | 2025-04-16 | Unknown | Closed |
12 | ![]() GPT-4.1 Standard Mode | 72.00 | 2025-04-14 | Unknown | Closed |
13 | ![]() GPT-4o(2024-11-20) Standard Mode | 71.00 | 2024-11-20 | Unknown | Closed |
14 | ![]() Claude Sonnet 3.7 Standard Mode | 68.00 | 2025-02-25 | Unknown | Closed |
15 | ![]() GPT-4o mini Standard Mode | 64.00 | 2024-07-18 | Unknown | Closed |
16 | ![]() OpenAI o4 - mini Thinking Level · Medium | 64.00 | 2025-04-16 | Unknown | Closed |
17 | ![]() OpenAI o4 - mini Thinking Level · High | 64.00 | 2025-04-16 | Unknown | Closed |
18 | ![]() Claude 3.5 Sonnet New Standard Mode | 62.00 | 2024-10-22 | Unknown | Closed |
19 | ![]() OpenAI o3 Thinking Level · High | 60.00 | 2025-04-16 | Unknown | Closed |
20 | Grok 4 Standard Mode | 45.00 | 2025-07-10 | Unknown | Closed |
On the official page observed on 2026-08-12, Claude Fable 5 reached 89.0% country accuracy on the 100-image ACW set.