Anthropic is out with a new model called Close work 4.6an upgrade of the top model Opus 4.5 that was launched in November. The new release could add new capabilities to Anthropic’s Claude Code coding assistant, which is facing increasing competitive pressure from OpenAI’s Codex.
Anthropic says Opus 4.6 improves on its predecessor’s coding skills and planning, and perhaps most importantly, the ability to reason more clearly when processing large amounts of information. When Opus 4.6 supports Claude Code, the coding agent can understand larger codebases and make more thoughtful decisions about how and where to add new code, the company says.
More long-term memory
AI labs are busy building models with longer context windows, meaning the amount of information a model can consider for a given task. But models have often struggled to use that information effectively in their output, a limitation Anthropic acknowledges.
“We used to see things like, maybe the model gets lost halfway through, or details are forgotten,” Opus product manager Dianne Penn tells Fast Company. “I wouldn’t say Opus 4.6 is perfect – humans and other past models aren’t perfect – but we think the quality improvement is quite significant.”
Opus’ longer memory also allows it to work on complex tasks for longer periods of time, allowing Claude Code users to create teams of agents who work together on tasks. Anthropic also says the tool offers improved code review and debugging capabilities, allowing it to catch its own errors.
Opus 4.6 arrives as the use of AI coding tools continues to increase and competition between Anthropic and OpenAI for software developers increases. OpenAI’s Codex encryption tool recently launched as a standalone app, powered by the GPT-5.2 model, and has received mostly enthusiastic reviews from developers.
A model for daily work tasks
In addition to coding, the new Anthropic model is designed to improve performance of everyday work tasks, such as performing financial analysis, conducting research, and creating or using documents, spreadsheets, and presentations. Opus 4.6 will also power Anthropic’s universal work tool, CoWork, allowing it to multitask with minimal human supervision.
Anthropic says Opus 4.6 achieved top scores in several industry benchmark tests, achieving the highest results to date in multiple evaluations. These include Humanity’s Last Exam, a complex multidisciplinary reasoning test; Terminal-Bench 2.0, an evaluation of agentic coding; and GDPval-AA, which measures performance on economically valuable knowledge work tasks in financial, legal, and other areas. Anthropic also says that Opus 4.6 outperforms all other models on OpenAI’s BrowseComp, which measures a model’s ability to find hard-to-find information online.
Anthropic says the Opus 4.6 model is available to developers using Claude Code for the same price per million tokens as Opus 4.5. The new model is now the standard for Claude Code Pro subscribers and available as an option for all other subscribers.
#Anthropics #Claude #Opus #aims #larger #codebases