Anthropic Releases Claude Opus 4.5 Model

Anthropic announced the release of Claude Opus 4.5 on Monday, unveiling its most advanced large language model to date as part of the company's 4.5 series. The new iteration, designed primarily for coding, office automation and complex agentic tasks, completes the lineup following the September launch of the midrange Sonnet 4.5 and the October debut of the entry-level Haiku 4.5. Available immediately through Anthropic's apps, desktop client and application programming interfaces, Opus 4.5 targets professionals seeking to streamline demanding workflows.

The model excels in areas such as programming and data analysis, achieving state-of-the-art results on key benchmarks. It scored over 80 percent on the verified SWEBench coding test, the first model to reach that threshold, and performed strongly on Terminalbench for tool use, Tau2bench and GPQA Diamond for problem-solving. Anthropic stated that Opus 4.5 requires fewer computational resources for similar tasks compared to predecessors, using fewer tokens while delivering higher efficiency. Internal testers reported that the model handles ambiguity effectively, reasoning through tradeoffs independently and resolving complex bugs...