QuoteBench: leaderboard
Metric: Task success (%; exact final-state validation over 56 one-shot Bash tasks in 14 operation families, 14 benign controls and 42 hostile payloads; disclosed-boundary generation contract with the reply replayed through one added nested shell parser; each model at its best-observed effort setting, one stored trial). Source: arxiv.org. Saturation forecast: Around December 2026. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 (xHigh) | 100 |
| 2 | Claude Fable 5 (Max) | 100 |
| 3 | Claude Opus 5 (xHigh) | 100 |
| 4 | GPT-5.6 Sol (High) | 98.2 |
| 5 | Claude Opus 4.8 (Max) | 94.6 |
| 6 | Gemini 3.1 Pro (Preview) (Low) | 91.1 |
| 7 | Gemini 3.5 Flash (Medium) | 64.3 |
| 8 | Claude Sonnet 4.6 (High) | 64.3 |
| 9 | Qwen 3.5 27B (Non-reasoning) | 30.4 |
| 10 | Qwen 3.5 4B | 21.4 |
| 11 | Qwen 3.5 9B (Non-reasoning) | 17.9 |
| 12 | Gemini 3.1 Flash Lite (Preview) | 14.3 |
Interactive version: theaggregate.ai/benchmark?slug=quotebench · How It Works · Data refreshed daily, snapshot 2026-09-26.