Anthropic releases flagship language model Claude Opus 5.5

According to the developer, the model generates responses more than 30% faster than its predecessor Opus 5 and costs on average 40% less on typical tasks. The base API rates are $4 per 1 million input tokens and $20 per 1 million output tokens compared to the previous $5 and $25 respectively, and cache reading has decreased from $0.50 to $0.20.

The developer is targeting the new model at prolonged engineering and analytical tasks. According to the company's statements, during preliminary testing, the model performed the migration of a software project with 680 thousand lines in less than a day, and the verification and editing of a database with 200 thousand lines in less than three hours, while Opus 5 took more than 20 hours for this. In the Terminal-Bench 4.0 benchmark, the model scored 66.4%.

Claude Opus 5.5 is already available to users of Claude API, the Claude Platform, as well as through the cloud services AWS, Google Cloud, and Microsoft Azure. For high-speed generation, the Fast Mode has been launched with an increased cost - $8 per million input and $40 per million output tokens.