Introducing Claude Opus 5
Simon Willison · Simon Willison · 2026-07-24
Simon Willison surveys Anthropic's Claude Opus 5 launch, noting it leads the Artificial Analysis leaderboard above Fable 5, exhibits highly proactive agentic behavior, and was deliberately not trained on cyberattack exploitation despite strong vulnerability-finding capabilities.
Appears in
Extraction
Topics: claude-opus-5llm-releaseagentic-aicybersecurityanthropic
Claims
- Claude Opus 5 is currently leading the Artificial Analysis leaderboard, placing above Claude Fable 5.
- Opus 5 is priced the same as Opus 4.8 and offers a fast mode at twice the base cost.
- Opus 5 demonstrates highly proactive problem-solving, including writing its own computer vision pipeline when no tool was provided.
- Anthropic deliberately avoided training Opus 5 on cyber exploitation tasks, leaving it substantially behind Mythos 5 on turning vulnerabilities into material threats.
- Opus 5 is described as 'thoughtful and proactive' and approaches Claude Fable 5's intelligence at half the price.
Key quotes
Anthropic's description of it as a 'thoughtful and proactive model that comes close to the frontier intelligence of Claude Fable 5 at half the price' sounds promising.
Opus 5 responded by writing its own computer vision pipeline to pull the geometry from the raw pixels, then reconstructed the full machine part.
it remains substantially behind Mythos 5 on the exploitation of those vulnerabilities—that is, in turning vulnerabilities into material cyber threats.