BioMedHop (Direct Prompting) - Path-based Counting: leaderboard
The Path-based Counting family of BioMedHop (Tan et al., 2026): count the distinct diseases that satisfy a typed SPOKE metapath. The score is exact-count accuracy in percent after alias deduplication; higher is better. Direct prompting without retrieved evidence.
Metric: Accuracy (%; exact count of the distinct diseases that satisfy a typed metapath, after alias deduplication (CountEx); stratified evaluation subsets of the 10,045 instances; direct input-output prompting with no retrieved evidence (the paper's IO baseline), temperature 0, at most 1,024 output tokens). Source: arxiv.org. Saturation forecast: Around 2030. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 13.5 | #262 |
| 2 | GPT-4 Turbo | 12.5 | #1002 |
| 3 | Claude 3.7 Sonnet | 12 | #455 |
| 4 | Qwen 3 Next 80B A3B Instruct | 11 | #763 |
| 5 | Claude 3.5 Haiku | 10.2 | #1563 |
| 6 | MiMo-V2.5-Pro | 9.5 | #274 |
| 7 | GPT-4o Mini | 8.5 | #1627 |
| 8 | Qwen 3 4B | 6 | #2297 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=biomedhop-direct-prompting-path-based-counting · How It Works · Data refreshed daily, snapshot 2026-11-12.