Anthropic launched Claude Opus 5.5, the first model in its new Claude 5.5 family, with the company positioning it as a major upgrade for coding, knowledge work, and AI agents. Anthropic says the model delivers performance comparable to its higher-end Claude Fable 5.1 on most tasks while costing about 40% less to operate than Opus 5.
The launch also puts Opus 5.5 directly into competition with OpenAI’s GPT-6 Astra. Anthropic’s published tests show its new model ahead of Astra on some benchmarks and roughly matching it on others, although the companies used different configurations and testing conditions in some comparisons.
Coding is one of the main areas where Anthropic says Opus 5.5 has made significant progress. The company highlighted its ability to handle lengthy, multistep software tasks, including large codebase migrations and debugging work.
In one early-user example cited by Anthropic, Opus 5.5 completed a migration involving around 680,000 lines of code in less than a day. Another test found that the model successfully improved page-load performance across a web application in 39 out of 40 attempts.
On Anthropic’s reported FrontierCode benchmark, Opus 5.5 scored 54.4%, compared with 53.3% for GPT-6 Astra. On Terminal-Bench 4.0, it recorded 66.4%, compared with Astra’s 57.9% under the listed test conditions.
Anthropic is also highlighting efficiency as a major part of the release. Opus 5.5 is priced at USD 4 per million input tokens and USD 20 per million output tokens, compared with USD 5 and USD 25 respectively for Opus 5.
The company says the model costs around 40% less on typical workloads and generates responses more than 30% faster. Anthropic also claims Opus 5.5 can match GPT-6 Astra on Terminal-Bench for about 40% of Astra's cost per task.
Also Read: Claude Gets One-Window Upgrade as Anthropic Brings Cowork Into Chat
Anthropic says Opus 5.5 underwent external testing before launch and performed strongly on its automated behavioral audit, which covers nearly 2,000 simulated scenarios. The company has also added safeguards to reduce prompt injection risks and unintended actions.
The model is now available through Anthropic’s platforms and can also be accessed through Amazon Web Services, Google Cloud, and Microsoft Azure. Anthropic plans to release Claude Sonnet 5.5 and Haiku 5.5 in the coming weeks, expanding the new model family beyond its flagship Opus release.