Anthropic has introduced Claude Opus 4.5, a hybrid reasoning model for coding, agents, and computer use. The company said this new version of the Claude Opus model offers better vision, reasoning, and mathematics than predecessors.

Claude Opus 4.5 is meaningfully better at everyday tasks such as deep research and working with slides and spreadsheets, Anthropic said.

Claude Opus 4.5 was introduced November 24. It is available now in Anthropic’s consumer applications, its API, and on all three major cloud platforms including Amazon Web Services, Google Cloud Platform, and Microsoft Azure. Developers can access the model by using claude-opus-4-5-20251101 via the Claude API.

According to Anthropic, testers of Claude Opus 4.5 found that the model handles ambiguity and reasons about tradeoffs without hand-holding, and that tests that proved near-impossible for Sonnet 4.5 only a few weeks ago were now within reach of Opus 4.5. Further, with Claude Opus 4.5, Anthropic has made substantial progress in robustness against prompt injection attacks, which smuggle in deceptive instructions to fool the model into harmful behavior, the company said.

Claude Opus 4.5 uses dramatically fewer tokens than its predecessors to reach similar or better outcomes, Anthropic said. Pricing for the new model is $5 per million tokens of input and $25 per million tokens of output.

Anthropic also announced updates to the Claude Developer Platform, Claude Code, and its consumer apps.