Firefox 148 Exploit Benchmark
- Entity ID:
ent-20260419-g1a0000000a6 - Type:
dataset - Scope:
shared - Status:
active - Aliases: Firefox 147 JS engine CVE benchmark, Mythos Firefox benchmark
Description
Internal Anthropic benchmark converting known Firefox 147 JS-engine vulnerabilities (all patched in Firefox 148) into working shell exploits. Opus 4.6 produced 2 working exploits across several hundred attempts (~0% success). Claude Mythos Preview produced 181 working exploits plus 29 additional achieving register control on the same benchmark - the most striking single capability-gap data point in the leak cycle.
Key claims
- Mythos Preview produces 181 working Firefox exploits vs Opus 4.6's 2
Relations
- Firefox 148 Exploit Benchmark --[supports]--> Claude Mythos Preview
- Firefox 148 Exploit Benchmark --[related_to]--> Fennec