HPR-Bench - Profile Summarization - Tag Coverage: leaderboard
Metric: Extracted-Tag Coverage (%; COVEx, share of profile tags recovered from the summary). Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 235B A22B 2507 (Thinking) | 96.38 |
| 2 | Qwen 3 235B A22B 2507 Instruct | 94.71 |
| 3 | GLM-5 (Thinking) | 94.22 |
| 4 | DeepSeek V3.2 (Thinking) | 93.41 |
| 5 | Kimi K2.5 (Thinking) | 93.32 |
| 6 | DeepSeek R1 0528 | 92.21 |
| 7 | Qwen 3.6 Plus (Thinking) | 90.46 |
| 8 | Qwen 3 8B (Non-reasoning) | 88.21 |
| 9 | Qwen 3 8B (Thinking) | 88.01 |
Interactive version: theaggregate.ai/benchmark?slug=hpr-bench-profile-summarization-tag-coverage · How It Works · Data refreshed daily, snapshot 2026-09-25.