MobilePA-Bench - Sub-agent Collaboration: leaderboard
Metric: Routing-and-handoff joint success (%). Source: arxiv.org. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 77.53 |
| 2 | Claude Fable 5 | 70.79 |
| 3 | Gemini 3.6 Flash | 66.29 |
| 4 | Kimi K3 | 62.92 |
| 5 | Claude Opus 5 | 62.92 |
| 6 | GLM-5.2 | 61.8 |
| 7 | Seed 2.1 Pro | 59.55 |
| 8 | Qwen 3.8 Max | 53.93 |
| 9 | GPT-5.5 | 51.69 |
| 10 | Claude Opus 4.8 | 50.56 |
| 11 | Qwen 3.7 Max | 50.56 |
| 12 | GPT-5.6 Sol | 49.44 |
| 13 | Kimi K2.6 | 43.82 |
Interactive version: theaggregate.ai/benchmark?slug=mobilepa-bench-sub-agent-collaboration · How It Works · Data refreshed daily, snapshot 2026-09-19.