TemporalBench - MIMIC T4: leaderboard
TemporalBench MIMIC clinical slice for T4 event-conditioned forecasting over patient measurements.
Metric: Accuracy (%). Source: huggingface.co. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | CAMEL (deepseek-chat) | 31.21 | |
| 2 | MetaGPT (deepseek-chat) | 30.5 | |
| 3 | Single LLM (gpt-4o) | 29.79 | |
| 4 | TimeClaw (deepseek-v4-pro) | 28.37 | |
| 5 | TimeCopilot (deepseek-chat) | 27.27 | |
| 6 | TimeCopilot (deepseek-v4-pro) | 27.08 | |
| 7 | MetaGPT (gpt-4o) | 25.53 | |
| 8 | AgentScope (deepseek-chat) | 24.82 | |
| 9 | CAMEL (gpt-4o) | 23.4 | |
| 10 | TimeSeries Scientist (deepseek-chat) | 23.4 |
Interactive version: theaggregate.ai/benchmark?slug=temporalbench-mimic-t4 · How It Works · Data refreshed daily, snapshot 2026-10-20.