Anthropic has released Claude Opus 5.5, the first model in its new 5.5 family, with lower API prices and a focus on long-running coding and agent work. The September 22 announcement positions the model between the previous Opus 5 and the company's higher-end Fable 5.1 tier.
Anthropic says Opus 5.5 performs at the level of Fable 5.1 on most tasks while costing about 40% less to run than Opus 5 on typical workloads at default settings. That figure is a company estimate, not a flat 40% cut to every price on the rate card.
Lower token prices, with a bigger cache discount
Standard input pricing falls from $5 to $4 per million tokens, and output pricing drops from $25 to $20. Cache reads fall more sharply, from $0.50 to $0.20 per million tokens. Because cached context makes up a large portion of many long agent sessions, Anthropic says the combined price cuts and lower token use produce the larger typical-workload savings.
The company also claims Opus 5.5 generates output more than 30% faster than Opus 5. A separate fast mode can run up to 2.5 times faster, but doubles standard Opus 5.5 prices to $8 per million input tokens and $40 per million output tokens.
Stronger model, tighter safeguards
Anthropic reports improvements across agentic coding, computer use and professional knowledge work. Those benchmark and early-customer results come from the company and its selected evaluators, so real-world performance will depend on the task, effort setting and tool setup.
Anthropic says external groups including Frontier Design and METR tested the model before release. Opus 5.5 also receives safeguards similar to Fable 5.1 for cybersecurity and biology, and it can transparently route some restricted work to another model. The company says evaluation is still imperfect and notes that the model may recognize when it is being tested.
Opus 5.5 is available in Claude for paid individual and business plans, through the Claude API and on Amazon Web Services, Google Cloud and Microsoft Azure under the model name claude-opus-5-5.
