MetaGAI - Model Card Quality: leaderboard
Metric: Qualitative average (1-5): mean judge Likert score of model cards generated zero-shot from the paper text alone, rated on faithfulness, relevance, accuracy, consistency and usefulness by GPT-OSS-120B, Llama-3.3-70B-Instruct and Qwen3-235B-A22B-2507 judges, on a stratified sample of 250 MetaGAI entries (2,541 verified paper, repository and Hugging Face triplets); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 30B A3B Instruct | 4.55 |
| 2 | GPT-5 Mini | 4.28 |
| 3 | Gemini 2.5 Flash Lite | 4.03 |
| 4 | Gemini 2.5 Flash | 4 |
| 5 | Gemma 3 27B (IT) | 3.94 |
| 6 | nemotron-3-nano-30B-a3B | 3.74 |
| 7 | Mistral Small 3.2 | 3.63 |
| 8 | Qwen 3 32B | 3.6 |
| 9 | GPT-OSS-20B | 3.38 |
| 10 | GPT-5 Nano | 3.23 |
Interactive version: theaggregate.ai/benchmark?slug=metagai-model-card-quality · How It Works · Data refreshed daily, snapshot 2026-10-07.