Anthropic Releases Claude Opus 5 for Coding and Office Tasks

Anthropic released its new Claude Opus 5 model this week, positioning it as a versatile AI capable of handling both complex coding tasks and general office work. The company claims Opus 5 delivers performance comparable to its more advanced Fable 5 model but at half the price. A key feature highlighted by Anthropic is Opus 5's ability to navigate complex problems with minimal user interaction, self-correct errors, and avoid getting stuck, representing an improvement over its predecessor, Opus 4.8, particularly in intricate coding scenarios. Anthropic describes Opus 5 as its most capable model to date for general knowledge work.

The introduction of Opus 5 is part of Anthropic's strategy to broaden Claude's applications beyond chat and code generation into a wider array of professional domains. These include design, marketing, accounting, and the analysis of extensive document sets. The model's performance is notable for its closeness to Fable 5, Anthropic's most advanced publicly available model, which excels in software writing, long-term projects with minimal supervision, and challenging assignments. Concerns were reportedly raised by the U.S. government regarding Fable 5's potential to identify and exploit software vulnerabilities for cyberattacks.

Opus 5 is priced at $5 per million input tokens and $25 per million output tokens. This pricing structure allows API customers to utilize Opus 5 for standard coding or office tasks and then switch to the more powerful, albeit more expensive, Fable 5 model for highly demanding work. Anthropic also asserts that Opus 5 demonstrates its most significant advancements for scientific research among its generally available models, citing enhanced performance in biology-related tasks compared to Opus 4.8, which could offer substantial benefits to researchers.