MedConclusion - Formality Similarity: leaderboard
Metric: Formality similarity (judge score 0-100): the model writes a formal academic conclusion (prompt A, no length or style constraint) from the non-conclusion sections of a PubMed structured abstract; a GPT-5.4-mini judge scores it 0-100 against the author-written conclusion, averaged over a random 30K-abstract subset of MedConclusion; temperature 0, at most 1,024 new tokens; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.4 (Non-reasoning) | 89.8 |
| 2 | Gemini 3.1 Pro (Preview) | 89.49 |
| 3 | Gemma 3 27B (IT) | 89.36 |
| 4 | GLM-4.6V | 88.87 |
| 5 | MiniMax-M2.1 | 88.83 |
| 6 | Kimi K2 (Thinking) | 88.62 |
| 7 | DeepSeek V3.2 (Non-reasoning) | 88.59 |
| 8 | Qwen 3 4B 2507 Instruct | 88.47 |
| 9 | Gemma 2 9B (IT) | 88.41 |
| 10 | Llama 3.1 8B Instruct | 88.03 |
| 11 | Qwen 2.5 VL 7B Instruct | 87.34 |
| 12 | Qwen 2.5 7B Instruct | 86.6 |
| 13 | Llama 3.2 1B Instruct | 78.35 |
| 14 | DeepSeek R1 | 75.91 |
Interactive version: theaggregate.ai/benchmark?slug=medconclusion-formality-similarity · How It Works · Data refreshed daily, snapshot 2026-10-07.