CodeAssay (Security-Focused Prompt): leaderboard

Metric: Final hidden-test correctness (%; share of 185 audited Python tasks in ten software-engineering categories whose scored solution passes every hidden test; one generation from the description, signature and a public-test excerpt plus at most one repair guided by public-test failures; provider-default reasoning, temperature 0 where accepted, 16,384-token output limit; security-focused system prompt that adds input validation, no eval/exec, secrets-module randomness, sanitized external data and error handling that exposes no internals). Source: arxiv.org. Saturation forecast: Estimated already saturated. 7 models tracked.

Top models

#ModelScore
1GPT-5.6 Sol98.4
2Claude Sonnet 597.8
3Claude Sonnet 4.595.7
4Claude Haiku 4.594.6
5Gemini 2.5 Flash91.4
6GPT-4o89.7
7GPT-4o Mini75.1

Interactive version: theaggregate.ai/benchmark?slug=codeassay-security-focused-prompt · How It Works · Data refreshed daily, snapshot 2026-09-29.