ShatterMed-QA (English, Hard): leaderboard
Metric: Accuracy (%) on the 1,692 English Hard multiple-choice clinical vignettes synthesized from a hub-pruned (k-shattered) medical knowledge graph, with the bridging entity masked and a hard distractor derived from its sibling node; zero-shot, closed-book; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4 Fast (Reasoning) | 97.87 | #242 (Grok 4 Fast) |
| 2 | Grok 4.1 Fast (Reasoning) | 97.58 | #208 (Grok 4.1 Fast) |
| 3 | GPT-4.1 Mini | 97.28 | #346 |
| 4 | GPT-5 Mini | 96.81 | #176 |
| 5 | Grok 4.1 Fast (Non-reasoning) | 96.1 | #208 (Grok 4.1 Fast) |
| 6 | Grok 4 Fast (Non-reasoning) | 96.04 | #242 (Grok 4 Fast) |
| 7 | GPT-4.1 Nano | 93.85 | #716 |
| 8 | GPT-5 Nano | 89.83 | #415 |
| 9 | Qwen3-14B-Base | 86.47 | #610 |
| 10 | Falcon3-10B-Base | 83.7 | #1191 |
| 11 | InternLM3-8B-Instruct | 78.44 | #847 |
| 12 | Llama 3.1 8B | 72.3 | #1139 |
| 13 | c4ai-command-r7B-12-2024 | 71.59 | #1145 |
| 14 | Yi-1.5-9B | 69.17 | #1230 |
| 15 | Gemma 2 9B | 55.82 | #940 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=shattermed-qa-english-hard · How It Works · Data refreshed daily, snapshot 2026-10-11.