NYU CTF Bench - D-CIPHER: leaderboard
Metric: Challenges solved (%; 200 challenges, $3 budget). Source: arxiv.org. Saturation forecast: Estimated already saturated. 5 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude 3.5 Sonnet (20241022) | 19 |
| 2 | GPT-4o (2024-11-20) | 10.5 |
| 3 | GPT-4 Turbo | 6.5 |
| 4 | Llama 3.1 405B Instruct | 3 |
| 5 | Gemini 1.5 Flash | 2.5 |
Interactive version: theaggregate.ai/benchmark?slug=nyu-ctf-bench-d-cipher · How It Works · Data refreshed daily, snapshot 2026-10-09.