TrainingPhoneBuddy
Training open phone-use models with real-app RL and PhoneWorld-style mock-app training, showing that realism and scalable verified interaction are complementary.
Open Phone-Use Agents
PhoneBuddy: Training Open Models for Agentic Phone Use
PhoneBuddy studies how to train open phone-use agents with both real-app execution and scalable mock-app environments. The core result is simple: real-app RL provides realism, while PhoneWorld-style mock-app training adds resettable and automatically verified interaction signal.
4Bopen phone-use model line
150real-phone human-evaluation tasks
83.2%AndroidWorld success rate after Real+Mock RL
+5.0average gain over real-app RL alone
Phone-Agent Research Line
PhoneBuddy sits in a broader phone-agent stack: environments for scalable interaction, training recipes for open models, runtime harnesses for real execution, and deployment boundaries for privacy and safety.
Phone-Agent Stack
EnvironmentPhoneWorld
A scalable pipeline that turns real GUI trajectories and screenshots into controllable phone-use environments, executable tasks, verifiers, and training rollouts.
RuntimePhoneHarness
A mixed-action phone-agent harness and benchmark that routes across CLI, GUI, and MCP tools with trace-backed verification.
PrivacyPhonePrivacy
A verifiable benchmark for privacy behavior in mobile agents, auditing permissioned access, minimal disclosure, and user-controlled memory.
SafetyPhoneSafety
Safety evaluation for phone-use agents, separating true safety behavior from simple incapability and checking risky mobile side effects.
Results
PhoneBuddy-4B-Real+Mock improves the same open-model line across real-phone and AndroidWorld settings. The visual summary below separates main capability comparison from the cross-app limitation.
Capability Profile
The radar compares PhoneBuddy-4B-Real+Mock against GPT-5.4, Gemini 3.1 Pro, and Seed 2.0 across the same five axes. PhoneBuddy is strongest on AndroidWorld and single-app tasks, while cross-app transfer remains the visible weak axis.
PhoneBuddy
GPT-5.4
Gemini
Seed
Average Success Rate
PhoneBuddySFT
GPT-5.4PhoneBuddyReal
Seed 2.0PhoneBuddyReal+Mock
Gemini3.1 Pro
Mixed real+mock training is the strongest PhoneBuddy checkpoint and closes much of the gap to larger proprietary agents.
AndroidWorld
PhoneBuddySFT
GPT-5.4Seed 2.0PhoneBuddyReal
Gemini3.1 Pro
PhoneBuddyReal+Mock
AndroidWorld shows a clean progression from SFT to real-app RL to mixed real+mock RL.
Single-App Tasks
PhoneBuddySFT
Seed 2.0GPT-5.4Gemini3.1 Pro
PhoneBuddyReal
PhoneBuddyReal+Mock
GPT and Gemini are shown separately; both score 50.0 on this slice.
WeChat Mini-App Tasks
GPT-5.4PhoneBuddyReal
PhoneBuddySFT
PhoneBuddyReal+Mock
Gemini3.1 Pro
Seed 2.0Mock-app practice helps recover mini-app reliability after real-only training.
| Model | Single-App | Cross-App | WeChat Mini-App | AndroidWorld | Avg. |
|---|---|---|---|---|---|
| PhoneBuddy-4B-SFT | 34.0 | 22.0 | 54.0 | 60.3 | 42.6 |
| PhoneBuddy-4B-Real | 54.0 | 20.0 | 48.0 | 77.2 | 49.8 |
| PhoneBuddy-4B-Real+Mock | 62.0 | 18.0 | 56.0 | 83.2 | 54.8 |
Evaluation covers real-phone Single-App, Cross-App, WeChat Mini-App tasks, plus AndroidWorld.
Delta view of real-app RL over SFT and mixed real+mock RL over real-app RL.
Method
PhoneBuddy compares a shared SFT checkpoint, a real-app RL checkpoint, and a mixed real+mock RL checkpoint under the same backbone, action interface, and evaluation protocol.
Real-App Environment
Real devices and authentic apps expose account state, app logic, timing variation, permission flows, and real side effects.
PhoneWorld
Runnable mock apps reconstructed from real GUI usage structure provide resettable training tasks and automatic verification.
Real + Mock RL
The final branch keeps real execution in the loop while adding scalable mock-app interaction for broader and cheaper training signal.
Real and Mock Environments
The real-app environment anchors training to authentic behavior; the mock-app environment contributes scale, reset, and automatic checking.
Qualitative Examples
Representative successful trajectories show structured workflows where mock-app training improves execution reliability.
Limitation & Future Work: Cross-App Generalization
PhoneWorld-style mock tasks currently emphasize single-app interaction rather than cross-app handoff. The ablation therefore compares only PhoneBuddy checkpoints and shows that mock-domain coverage matters for cross-app workflows.
PhoneBuddyReal
PhoneBuddySFT
PhoneBuddyReal+Mock