QuarkMedBench - Highlight: leaderboard
Metric: Highlight score rate (%): share of the Highlight rubric points (beneficial extended recommendations beyond the core facts) that the answer earns, on a stratified sample of 2,000 real Chinese online medical and health queries (single- and multi-turn), answers scored against query-specific rubrics (Essential, Important, Highlight and Pitfall items) by a Qwen3-235B judge, without an output length limit; higher is better. Source: arxiv.org. Saturation forecast: Around 2032. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 30B A3B 2507 (Thinking) | 49.11 | #366 (Qwen 3 30B A3B 2507) |
| 2 | Gemini 2.5 Pro | 45.01 | #145 |
| 3 | DeepSeek V3.2 | 41.86 | #198 |
| 4 | GPT-5.1 | 40.37 | #131 |
| 5 | Qwen 3 235B A22B (Non-reasoning) | 40.12 | #304 (Qwen 3 235B A22B) |
| 6 | DeepSeek R1 0528 | 38.79 | #217 |
| 7 | Gemini 3 Pro | 38.75 | #77 |
| 8 | Qwen 3 32B (Non-reasoning) | 38.3 | #424 (Qwen 3 32B) |
| 9 | GPT-5.2 | 33.76 | #105 |
| 10 | Seed-1.6 | 32.3 | #257 |
| 11 | Qwen 3 30B A3B (Non-reasoning) | 31.38 | #488 (Qwen 3 30B A3B) |
| 12 | Qwen 3 8B (Non-reasoning) | 30.82 | #667 (Qwen 3 8B) |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=quarkmedbench-highlight · How It Works · Data refreshed daily, snapshot 2026-10-11.