GeoBench

Guessing where a photo was taken: country accuracy.

Results

#	Model	Provider	Result	Date
1	Gemini 3 Flash Preview	Google	88.0%	2026-11-11
2	Gemini 2.5 Pro Preview 06-05	Google	86.0%	2026-11-11
3	Gemini 3 Pro Preview	Google	84.0%	2026-11-11
4	GPT-5 Medium	OpenAI	81.0%	2026-11-11
5	o1 Medium	OpenAI	80.0%	2026-11-11
6	Gemini 2.0 Flash	Google	77.0%	2026-11-11
7	Gemini 2.5 Flash Preview 05-20	Google	76.0%	2026-11-11
8	Gemini 1.5 Flash (Sep 2024)	Google	76.0%	2026-11-11
9	Claude Opus 4.5	Anthropic	75.0%	2026-11-11
10	o3 Medium	OpenAI	74.0%	2026-11-11
11	Gemini 2.5 Flash (Apr 2025)	Google	73.0%	2026-11-11
12	GPT-4.1	OpenAI	72.0%	2026-11-11
13	GPT-4o (2024-11-20)	OpenAI	71.0%	2026-11-11
14	Claude Sonnet 3.7	Anthropic	68.0%	2026-11-11
15	Claude Sonnet 3.7 Thinking	Anthropic	65.0%	2026-11-11
16	GPT-4o mini	OpenAI	64.0%	2026-11-11
17	o4-mini Medium	OpenAI	64.0%	2026-11-11
18	o4-mini High	OpenAI	64.0%	2026-11-11
19	Claude Sonnet 3.5 v2	Anthropic	62.0%	2026-11-11
20	Qwen2.5-VL 72B Instruct	Alibaba	62.0%	2026-11-11
21	o3 High	OpenAI	60.0%	2026-11-11
22	Llama-3.2-90B-Vision-Instruct	Meta	52.0%	2026-11-11
23	Gemma 3 27B IT	Google	52.0%	2026-11-11
24	Llama 4 Maverick	Meta	52.0%	2026-11-11
25	Claude Opus 4 Thinking	Anthropic	49.0%	2026-11-11
26	Grok 4	xAI	45.0%	2026-11-11
27	Pixtral 12B	Mistral	37.0%	2026-11-11
28	Claude Sonnet 4	Anthropic	37.0%	2026-11-11
29	Claude Haiku 3.5	Anthropic	34.0%	2026-11-11
30	Claude Sonnet 4 Thinking	Anthropic	32.0%	2026-11-11