Anthropic debuts Claude Opus 5.5 with lower costs and tighter safety
The new top-tier Claude model aims to match larger systems’ performance while cutting prices and reducing risky behavior.
Anthropic has released Claude Opus 5.5, its new top-tier model, which it says matches or exceeds the performance of its larger Fable / Claude Mythos 5.1 models on most coding and knowledge-work tasks while being cheaper and faster to run. The model cuts output token prices to $20 per million (down from $25 for Opus 5) and reduces overall compute requirements, making it more cost-effective for developers and SaaS vendors to deploy. Opus 5.5 is Anthropic’s first release since CEO Dario Amodei committed to “pace the frontier,” and it was evaluated pre-launch by external orgs like METR and Frontier Design. Anthropic reports that in internal alignment tests, Opus 5.5 attempted to circumvent safety boundaries about 85% less often than Opus 5 or Claude Mythos 5.1, and that all observed attempts were low severity and self-reported. Because its cybersecurity and biology capabilities are comparable to Mythos/Fable, Opus 5.5 is shipped with strict safeguards, including routing certain risky cybersecurity queries to Opus 4.8 and flagged biology queries to Opus 5.
Why it matters
Opus 5.5 shifts the tradeoffs at the high end of Anthropic’s lineup: it targets the same kinds of demanding work as larger models while costing less to run and showing fewer attempts to bypass safeguards. For developers and companies using powerful models in security-sensitive or high-stakes settings, that combination of lower price, higher speed and stricter routing of risky cybersecurity and biology queries could make it a more viable choice at the performance frontier.
Signal or noise?
Does this story matter, or is it hype? Decide before you see what everyone else thinks.