MedBench v5 - DeepResearch: leaderboard
Metric: Environment score (0-100) on DeepResearch (long-horizon medical research planning and report synthesis over PubMed and web sources; semantic recall and agent-as-a-judge); higher is better. Source: arxiv.org. Saturation forecast: Around July 2027. 4 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Seed 2.0 Pro | 62.58 |
| 2 | GPT-5.5 | 53.26 |
| 3 | Kimi K2.6 | 41.16 |
Interactive version: theaggregate.ai/benchmark?slug=medbench-v5-deepresearch · How It Works · Data refreshed daily, snapshot 2026-09-29.