Anthropic's newest flagship model claims Fable 5.1 level output at a fraction of Opus 5's price.
The company released Claude Opus 5.5 today as the first entry in a new 5.5 lineup. Anthropic says it matches Claude Fable 5.1 on most tasks while costing 40 percent less to run than Opus 5, and it generates responses over 30 percent faster. On benchmarks Anthropic shared, Opus 5.5 outperformed Fable 5.1 on agentic coding, knowledge work, computer use, chart reading, and multidisciplinary reasoning. Anthropic also says the model writes more plainly, leads with the important information, and sticks closer to whatever style rules it's given.
For anyone paying by the token, a flagship-adjacent model at a lower price than the previous flagship is the real story here, not the benchmark charts. But Anthropic's own safety framing raises a question it doesn't answer: the company says Opus 5.5 poses biology and cybersecurity risks comparable to Mythos 5.1, yet it's shipping with the lighter safeguards built for Fable 5.1 rather than Mythos 5.1's stricter ones. That's a mismatch between the risk Anthropic says the model carries and the guardrails placed around it.
Anthropic doesn't explain the gap, and until it does, calling this a major step up in safety is a claim worth reading twice.