PerceptionBench - Localization — leaderboard
Metric: Localization (%). Source: huggingface.co. 16 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Sol | 76.7 |
| 2 | Claude Fable 5 | 70.4 |
| 3 | Kimi K3 | 70.3 |
| 4 | GPT-5.5 | 65.8 |
| 5 | Claude Opus 4.8 | 58.8 |
| 6 | Gemini 3.1 Pro (Preview) | 52.7 |
| 7 | Qwen 3.7 Plus | 52.7 |
| 8 | Gemini 3.5 Flash | 50.6 |
| 9 | Seed 2.1 Pro | 50 |
| 10 | Qwen 3.5 397B A17B | 46.7 |
| 11 | Kimi K2.6 | 45.2 |
| 12 | Gemma 4 31B | 44.9 |
| 13 | GLM-5V Turbo | 43.9 |
| 14 | Grok 4.5 | 39.7 |
| 15 | MiniMax-M3 | 33.3 |
Interactive version: theaggregate.ai/benchmark?slug=perceptionbench-localization · How It Works · Data refreshed daily, snapshot 2026-08-05.