MobilePA-Bench - Basic Tool Use: leaderboard
Aggregate accuracy of a mobile planning agent calling tools in MobilePA-Bench's stateful sandbox with strict ordering, permission limits and runtime errors (MobilePA-Bench, arXiv 2608.23035v2, Table 3). Percent; higher is better.
Metric: Aggregate accuracy (%). Source: arxiv.org. Saturation forecast: Around February 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 | 83.85 | #16 |
| 2 | Claude Fable 5 | 83.37 | #20 |
| 3 | Gemini 3.1 Pro (Preview) | 80.58 | #64 |
| 4 | Claude Opus 4.8 | 79.04 | #39 |
| 5 | Gemini 3.6 Flash | 78.65 | #57 |
| 6 | Qwen 3.8 Max | 77.88 | #34 |
| 7 | Kimi K3 | 77.4 | #35 |
| 8 | Qwen 3.7 Max | 76.54 | #85 |
| 9 | GLM-5.2 | 76.06 | #80 |
| 10 | Seed 2.1 Pro | 72.98 | #73 |
| 11 | Kimi K2.6 | 70.38 | #113 |
| 12 | GPT-5.6 Sol | 69.81 | #22 |
| 13 | GPT-5.5 | 68.94 | #33 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mobilepa-bench-basic-tool-use · How It Works · Data refreshed daily, snapshot 2026-11-05.