PerceptRL Agentic - RealWorldCounting: leaderboard
PerceptRL tests visual understanding and spatial reasoning, with separate results for single-turn answers and agents using tools. Accuracy (%).
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Around 2028. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Doubao Seed 2.0 Mini | 33.7 | #160 |
| 2 | Gemini 3 Flash | 33.2 | #103 |
| 3 | Claude Opus 4.8 | 29.2 | #39 |
| 4 | Kimi K2.6 | 25 | #111 |
| 5 | Grok 4.5 | 23.9 | #53 |
| 6 | Qwen 3.6 35B A3B | 19.4 | #233 |
| 7 | Qwen 3.5 Flash | 17.7 | #235 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=perceptrl-agentic-realworldcounting · How It Works · Data refreshed daily, snapshot 2026-10-29.