CLIR-Bench - Time-Series Summarization: leaderboard
Metric: Accuracy (%) on clinical time-series summarization, 4-option multiple-choice questions over serialized irregular ICU time series (Full-TS input), chance 25; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.4 Mini | 93.33 |
| 2 | Gemini 2.5 Flash | 91.67 |
| 3 | Qwen 3.6 27B | 74 |
| 4 | Kimi K2.6 | 58.33 |
| 5 | Qwen 3.5 9B | 30 |
| 6 | DeepSeek V4 Flash | 25 |
| 7 | Qwen 3.5 4B | 24.83 |
| 8 | Gemma 3 27B (IT) | 21.67 |
| 9 | Gemma 4 12B (IT) | 17.67 |
Interactive version: theaggregate.ai/benchmark?slug=clir-bench-time-series-summarization · How It Works · Data refreshed daily, snapshot 2026-09-29.