Do Coding Agents Understand Least-Privilege Au: leaderboard
Metric: TSR (self-reported). Source: benchmarklist.com. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 85.8 |
| 2 | GPT-5 | 76.7 |
| 3 | Kimi K2.5 | 70 |
| 4 | GPT-5.3 Codex | 65.8 |
| 5 | Qwen3 Coder | 63.3 |
| 6 | Claude Opus 4.6 | 61.5 |
| 7 | GPT-5.4 | 61.1 |
| 8 | MiniMax-M2.7 | 48.3 |
| 9 | Qwen 3.5 397B A17B | 45.6 |
Interactive version: theaggregate.ai/benchmark?slug=do-coding-agents-understand-least-privilege-au · How It Works · Data refreshed daily, snapshot 2026-09-05.