KDR-Bench - Key Point Coverage: leaderboard
Use of structured knowledge on KDR-Bench (2026), a deep research benchmark of 41 expert-level questions: the share of 261 expert-reviewed key points, each derived from a specific statistics table, that the generated report covers, judged by DeepSeek-V3.2. Percent; higher is better.
Metric: Key point coverage (%): share of each question’s annotated table-grounded key points that its report covers (261 key points in total), on the 41 expert-level research questions of KDR-Bench in nine domains backed by 1,252 Statista tables, each report judged by DeepSeek-V3.2; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 9 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro Deep Research | 58.3 | |
| 2 | MiniMax-M2 | 50.5 | #689 |
| 3 | Perplexity Deep Research | 49.8 | |
| 4 | Grok 4 DeeperSearch | 49.5 | |
| 5 | Hunyuan 2.0 | 46.7 | #690 |
| 6 | Deep Research | 44.1 | |
| 7 | Qwen 3 Max | 43.5 | #415 |
| 8 | Tongyi DeepResearch 30B A3B | 41.2 | |
| 9 | GLM-4.6 | 40.1 | #498 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=kdr-bench-key-point-coverage · How It Works · Data refreshed daily, snapshot 2026-11-21.