Firefox 147 Exploit Development (Any Progress)
Safety · 2026-07-24
Share of Firefox 147 exploit-development trials reaching at least the partial-progress grade of 0.5, which requires register control, rather than a full working exploit. This is the more permissive of the benchmark's two reported rates.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Mythos 5 | 90.0 |
| Claude Opus 5 | 87.2 |
| Opus 4.8 | 68.8 |