Anthropic

Claude Mythos Preview

A frontier model from Anthropic with powerful cybersecurity skills for both defensive and offensive applications.

Our editorial assessment as of Jun 4, 2026, based on the public sources listed below.

Overall rank

#13

Tier

Score

87.0

Main category

Coding

Strengths

Demonstrates capabilities in software engineering and reasoning substantially beyond previous models.
Can be used for defensive security purposes like finding and fixing vulnerabilities.
Has identified thousands of high-severity vulnerabilities in existing software.

The 'Preview' status suggests it may not be in full general availability or could be unstable.
Its powerful offensive capabilities raise concerns about potential misuse if not released carefully.