PreScience (Contribution Generation): leaderboard
Metric: LACER score (1 to 10): a gpt-5-2025-08-07 judge rates the conceptual similarity of the generated title and abstract to the real paper's, calibrated with interpolated demonstrations (1 = a related but distinct prior paper, 10 = a paraphrase); input: the target paper's highly influential references; target papers from the October 2024 to September 2025 test period of recent AI research; higher is better. Source: arxiv.org. Saturation forecast: Around December 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 5.64 | #91 |
| 2 | GPT-5.2 | 5.6 | #105 |
| 3 | O3 | 5.49 | #121 |
| 4 | Claude Opus 4.5 | 5.04 | #79 |
| 5 | Claude Sonnet 4.5 | 5.03 | #138 |
| 6 | GPT-4o | 4.71 | #333 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=prescience-contribution-generation · How It Works · Data refreshed daily, snapshot 2026-10-11.