RedCode - Exec Bash (ReAct): leaderboard
Metric: Attack success rate (%). Source: redcode-agent.github.io. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 3.1 8B Instruct | 59.83 |
| 2 | GPT-4 Preview (1106) | 61.96 |
| 3 | Llama 3 8B Instruct | 62.25 |
| 4 | Claude 3.5 Sonnet (20240620) | 62.67 |
| 5 | GPT-3.5 Turbo (1106) | 70.38 |
| 6 | CodeQwen1.5-7B Chat | 71.92 |
| 7 | GPT-4o (2024-05-13) | 72.83 |
| 8 | deepseek-coder-6.7B-instruct | 73.17 |
| 9 | Llama 3.1 70B Instruct | 74.38 |
Interactive version: theaggregate.ai/benchmark?slug=redcode-exec-bash-react · How It Works · Data refreshed daily, snapshot 2026-10-09.