World Agent: leaderboard
Metric: Maintenance completion score (%): share of a world's contracted events, structures and probes realized, mean per run over the nine-case common subset (27 runs per model, easy/hard/extreme tiers, PhyAgentOS harness, isolated LLM judge plus programmatic audit); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.8 Max (0902) | 81.3 |
| 2 | GLM-5.3 Flash | 75.2 |
| 3 | DeepSeek V4.1 Flash | 74.4 |
| 4 | Qwen 3.8 Flash | 73.5 |
| 5 | GPT-5.6 Luna | 73.4 |
| 6 | Kimi K3 | 70.9 |
| 7 | Gemini 3.8 Flash | 70.4 |
| 8 | GLM-5.3 | 69.2 |
Interactive version: theaggregate.ai/benchmark?slug=world-agent · How It Works · Data refreshed daily, snapshot 2026-09-29.