Anthropic launches Claude Opus 4 and Sonnet 4, Claude Code reaches general availability
Anthropic's two new models focus on long-running coding and agentic work, while Claude Code moves from preview to a general release.
Anthropic has announced Claude Opus 4 and Claude Sonnet 4. According to the company, both are designed for long-running coding tasks and agentic tasks, meaning the model carries out a chain of steps itself instead of answering one question at a time. In the same announcement, Claude Code moved from preview status to a general release.
The notable point lies in how they are positioned: instead of being optimised for a single short answer, both models aim to keep the thread of work across many steps, read and modify source code, then carry on when more work is assigned. Claude Code brings that capability into the command line environment, where engineering teams already work with git, compilers and testing tools.
For businesses, the change reinforces the direction of putting AI assistants into the software development process rather than using them only for question and answer. A tool that runs in the command line can take part directly in the daily work of the engineering team.
What to keep in mind is that the more steps an autonomous model can complete, the higher the control requirements become. Engineering teams should define clearly which scope of changes is permitted, how results are reviewed and what the approval process is before letting a tool operate on a real source code repository.
Source: anthropic.com
