GeoGuess Bench launches to test models at GeoGuessr
The benchmark gives each model 210 photos from around the world and scores how close its location guesses are; Muse Glimmer 30B beat GPT-6 Astra at about 45x lower cost, Claude Opus 5.5 took first place, and open models did surprisingly well.
Introducing GeoGuess Bench! A benchmark to measure how good AI models are at playing GeoGuessr. We gave each model 210 photos from around the world, asked it to predict where each photo was taken, and scored it based on how close it got. Some surprising results: • Muse Glimmer 30B beat GPT 6 Astra while being ~45x cheaper • GLM 5.3 Flash matched GPT 6.1 Sol while being ~5x cheaper • Claude Opus 5.5 took the #1 spot, scoring even higher than Fable 5.1 Open models did surprisingly well overall. Check out the full leaderboard, Pareto frontier curve, and every model's guesses! - Site → http://geoguessbench.com - Code → http://github.com/Nutlope/geoguessbench