Vector Eval - InterCode-CTF: leaderboard
Vector Institute Inspect Evals run of InterCode-CTF, testing cybersecurity agents on capture-the-flag tasks that require terminal interaction.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Around January 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen3 Coder Next | 85.9 | #326 |
| 2 | Claude 3.5 Sonnet | 85.57 | #350 |
| 3 | O1 | 84.81 | #183 |
| 4 | O3 Mini | 82.78 | #272 |
| 5 | GPT-4o | 63.8 | #343 |
| 6 | Gemini 1.5 Pro | 52.91 | #445 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vector-eval-intercode-ctf · How It Works · Data refreshed daily, snapshot 2026-10-20.