Androidworld
AndroidWorld benchmark for evaluating autonomous agents on real Android apps with 116 tasks across 20 apps
- Domain
- rl-env
- License
- apache-2.0
- Published
- Mar 2026
Cite
Notes
Only stored in your browser.
Top score 26.7% by GPT-4.1 - 1 model reporting (1 frontier)
Top models
1
Related tools
1Implementations, trainers, datasets and scaffolds linked to this eval.
FAQ
- What is Androidworld?
- AndroidWorld benchmark for evaluating autonomous agents on real Android apps with 116 tasks across 20 apps
- What is the current top score on Androidworld?
- The top reported score is 26.7% by GPT-4.1, across 1 model reporting (1 from frontier labs).
- How can a model improve its Androidworld score?
- Tools linked to Androidworld on Sophon include Androidworld RL Env (Prime Community) - RL environments, datasets, and scaffolds that target this eval.
- What license is Androidworld under?
- Androidworld is available under apache-2.0.