CIPHER - Insecure Prompts: leaderboard

Metric: Vulnerability rate (% of 750 generations with at least one vulnerability; 150 insecure-guidance prompts x 5 samples; GPT-4o judge, majority of several passes, CIPHER taxonomy). Source: arxiv.org. Saturation forecast: Around 2035. 7 models tracked.

Top models

#ModelScore
1GPT-5.1 Codex65.3
2GPT-5.177.1
3Gemini 2.5 Flash77.9
4CodeLlama-7B-Instruct-hf84.5
5Claude Sonnet 4.588.9

Interactive version: theaggregate.ai/benchmark?slug=cipher-insecure-prompts · How It Works · Data refreshed daily, snapshot 2026-09-25.