List your company

Agentic Phone Use on AndroidWorld

83.2Task Success RatePhoneBuddy-4B-Real+Mock
54.91262.25669.676.944Jun 22, 2026

Evaluation results

MethodLinks
PhoneBuddy-4B-Real+MockParameters=4B, Training=mixed RL in real-app and mock-app2026.0683.2
Gemini 3.1 Pro2026.0680.2
PhoneBuddy-4B-RealParameters=4B, Training=real-app RL2026.0677.2
Seed 2.0 Pro2026.0671.5
GPT-5.42026.0670.7
PhoneBuddy-4B-SFTParameters=4B, Training=supervised fine-tuning baseline2026.0660.3
Claude Opus 4.72026.0656