SciImpact - Media: leaderboard

SciImpact pairs that differ in media attention, from paper titles and abstracts; the model sees two artifacts from the same field and picks the one with higher impact; pairwise accuracy in percent, chance 50, higher is better.

Metric: Pairwise prediction accuracy (%, times 100) on media-attention pairs, over SciImpact contrastive pairs (215,928 pairs over 19 fields, built from citation counts, best paper awards and Nobel prizes, patent citations, media attention, GitHub stars and Hugging Face downloads with impact thresholds): the model sees two same-field artifacts (title and abstract, README, dataset card or model card, truncated to 1,000 words) and names the one with higher future impact in a two-option forced choice parsed by exact match; the order is balanced so chance is 50%; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Claude Haiku 4.563.2#587
2O4 Mini62.7#372
3Qwen 2.5 14B62#1819
4GPT-4.1 Mini60.8#799
5Ministral 3 3B60.7#3606
6Qwen 2.5 7B60.4#2902
7Llama 3 8B59.4#5897
8Llama 3.1 8B58.9#5193
9Qwen 3 4B58.7#2379
10Llama 3.2 3B51.7#7835
11nemotron-3-nano-30B-a3B50.4#1902

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=sciimpact-media · How It Works · Data refreshed daily, snapshot 2026-11-22.