PerceptionBench - Contextual Integration — leaderboard
Metric: Contextual Integration (%). Source: huggingface.co. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 61.2 |
| 2 | Seed 2.1 Pro | 60.4 |
| 3 | GPT-5.6 Sol | 60 |
| 4 | Claude Fable 5 | 59.8 |
| 5 | GPT-5.5 | 58 |
| 6 | Gemini 3.5 Flash | 53.3 |
| 7 | Kimi K3 | 53.3 |
| 8 | Qwen 3.7 Plus | 52.2 |
| 9 | Qwen 3.5 397B A17B | 50.2 |
| 10 | Gemma 4 31B | 45.9 |
| 11 | Claude Opus 4.8 | 44.7 |
| 12 | Kimi K2.6 | 40.4 |
| 13 | Grok 4.5 | 39.6 |
| 14 | GLM-4.6V | 33.7 |
| 15 | GLM-5V Turbo | 32.9 |
Interactive version: theaggregate.ai/benchmark?slug=perceptionbench-contextual-integration · How It Works · Data refreshed daily, snapshot 2026-08-05.