Anthropic's new Claude Opus 5.5 costs less to run and writes more clearly

Anthropic released Opus 5.5 on 22 September. It is available on every paid Claude plan, and on the API it costs 20% less per token than Opus 5.

Anthropic released Claude Opus 5.5 on Tuesday 22 September, two months after Opus 5. Anthropic calls it the strongest model it has tested, and TechCrunch reports it beat the larger Fable model on many benchmarks and finished some tasks Fable could not.

In the Claude app, Opus 5.5 is available on the Pro, Max, Team and Enterprise plans, and Anthropic says those plans get higher five-hour usage limits with it. For anyone building on the API, the price is $4 per million input tokens and $20 per million output tokens, down from $5 and $25. Anthropic says it writes output 30% faster than Opus 5 and often reaches the same quality using about half the tokens, which it puts at around 40% cheaper on typical workloads. Sonnet 5.5 and Haiku 5.5 are due in the coming weeks.

For a midsize business, two parts of the release matter more than the benchmarks. Anthropic says the model is better at financial analysis and business research, and that its explanations are shorter and put the important information first. If your team uses Claude to summarise reports or draft analysis, that means less reading and less editing. If you pay for an automation that calls Opus through the API, the same job should now cost less.

Before you switch a working process over, run a handful of your real tasks through both models and compare the results side by side. Benchmarks are measured on someone else's work. If you have an automation pinned to an older model name, it will not move to Opus 5.5 by itself, so check which model it calls.

If Claude is already part of how your team works, the new model is worth a test this week. The smaller Sonnet and Haiku versions will follow.