Curated List: 7 Papers on Evaluating Mobile-Use Agents on Android
East-Muffin-6472 · reddit · 2026-11-22
A curated reading list for evaluating Android mobile-use agents, organized into two groups:
- Training + evaluation methodology: AppAgent, SeeClick (GUI grounding), AndroidWorld, AndroidLab
- Evaluation-only benchmarks: MobileWorld (MCP-augmented environments), AndroidDaily (real-world closed-source apps), AndroidReality (real-world robustness gap)
The author's goal is to build evals mirroring everyday tasks and benchmark models running locally on phones; they released an initial benchmark called AndroidLife and are soliciting feedback via papercase.dev.
Related event: Curated Reading List for Android Mobile Agent Evaluation(2 posts)→
More from coding & agent
- Users complain GPT-5.5 writes bloatware and duplicates functions — yacineMTB · 2026-11-22
- The only software that survives is what agents can use fast: frictionless onboarding as survival criterion — alexmacgregor__ · 2026-11-22
- Are Gemini 4 Neon fans actually using it in Antigravity? Coder asks — iScienceLuvr · 2026-11-22
- '2,000 lines from Claude, 10-second Codex review': devs fret over AI engineering quality — _jaydeepkarale · 2026-11-22
- Cairnmind: open-source MCP server enforcing read-before-write for agent memory — Zestyclose_Bus_8031 · 2026-11-22
- Team releases Part 2 of $102,725 AI episode, sharing 50,000-character Claude skill — azed_ai · 2026-11-22