PointerBench Specialists - Pro: leaderboard
PointerBench’s same 1,500 tasks, with 39 specialist configurations using their native input resolution, prompt and coordinate conversion. MVP and GUI-RC wrapper runs retain their distinct identities. Success rate (%).
Metric: Success rate (%). Source: arxiv.org. Saturation forecast: Around January 2027. 39 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 8B Instruct | 65.8 | #419 |
No result here: #5 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=pointerbench-specialists-pro · How It Works · Data refreshed daily, snapshot 2026-10-25.