CUSP Science: leaderboard
Forecasting benchmark for scientific progress where models answer cutoff-conditioned binary, multiple-choice, date, and free-response questions about future scientific milestones.
Metric: FRQ Pass (%). Source: seanwu25.github.io. Saturation forecast: Around January 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 60.3 | #79 |
| 2 | DeepSeek R1 | 20 | #251 |
| 3 | Claude Sonnet 4.5 | 14 | #142 |
| 4 | GPT-OSS-20B | 11.9 | #518 |
| 5 | GPT-4o | 3.9 | #343 |
| 6 | Llama 3.3 70B | 2.5 | #579 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=cusp-science · How It Works · Data refreshed daily, snapshot 2026-10-20.