WorldBench (Cultural Agents): leaderboard
Metric: Complete Task Success (%). Source: arxiv.org. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 49.2 |
| 2 | GPT-5 | 48.8 |
| 3 | Qwen 3 32B | 48 |
| 4 | GPT-4o | 38.6 |
| 5 | Gemini 3.5 Flash | 34.6 |
| 6 | Llama 3.3 70B | 24.1 |
| 7 | Llama 3.1 8B | 12.5 |
| 8 | Qwen 3 4B | 11.6 |
Interactive version: theaggregate.ai/benchmark?slug=worldbench-cultural-agents · How It Works · Data refreshed daily, snapshot 2026-09-20.