Citation Verifiability on 36 clinician-validated scenarios for rare neuromuscular disease pairs
577Count of PMIDsClaude Opus 4.6
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Claude Opus 4.6Mode=Citation-prompted2026.04 | 577 | 88 | 455 | 34 | |
| Claude Sonnet 4.6Mode=Citation-prompted2026.04 | 415 | 49 | 331 | 35 | |
| HEG-TKGMode=KG-grounded2026.04 | 203 | 203 | 0 | 0 | |
| DeepSeek-v3Mode=Citation-prompted2026.04 | 163 | 17 | 135 | 11 | |
| GPT-4.1Mode=Vanilla2026.04 | 2 | 0 | 2 | 0 | |
| GPT-5.4Mode=Vanilla2026.04 | 0 | — | — | — | |
| Claude Sonnet 4.6Mode=Vanilla2026.04 | 0 | — | — | — | |
| DeepSeek-v3Mode=Vanilla2026.04 | 0 | — | — | — |