TemporalBench - T4: leaderboard
TemporalBench T4 family measuring MCQ accuracy on harder event-conditioned forecasting and planning questions.
Metric: Accuracy (%). Source: huggingface.co. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | TimeClaw (deepseek-v4-pro) | 43.63 | |
| 2 | AgentScope (deepseek-chat) | 35.08 | |
| 3 | CAMEL (deepseek-chat) | 35.08 | |
| 4 | MetaGPT (deepseek-chat) | 34.56 | |
| 5 | Single LLM (deepseek-chat) | 28.79 | |
| 6 | TimeCopilot (deepseek-chat) | 28.35 | |
| 7 | CAMEL (gpt-4o) | 27.92 | |
| 8 | AgentScope (gpt-4o) | 27.57 | |
| 9 | Single LLM (gpt-4o) | 26.7 | |
| 10 | TimeSeries Scientist (deepseek-chat) | 26.7 |
Interactive version: theaggregate.ai/benchmark?slug=temporalbench-t4 · How It Works · Data refreshed daily, snapshot 2026-10-20.