Anthropic's Claude Opus 4.6 AI model improves reasoning, task handling
Anthropic has launched Claude Opus 4.6, bringing improvements to long-context reasoning, accuracy in coding, and extended task handling for developers and enterprise users
 "Anthropic's Claude Opus 4.6 AI model improves reasoning, task handling")
Anthropic has launched Claude Opus 4.6, an upgraded version of its AI model, focusing on improved reasoning, longer task handling, and stronger performance across coding and general knowledge work. According to Anthropic, the new model builds on Claude Opus 4.5 and is designed to handle complex workflows more reliably without losing context over extended sessions.
While coding remains a key area of improvement, Claude Opus 4.6 is not limited to software development. The company says the model is intended for a wider range of professional tasks, including research, financial analysis, and document creation, positioning it as a general-purpose system for knowledge workers and developers.
Coding and long-context capabilities
Anthropic says Claude Opus 4.6 plans coding tasks more carefully, performs better code reviews, and is more effective at identifying and correcting its own errors. The model is also said to work more reliably within large codebases, an area where earlier models could struggle over time.
One of the most notable updates is the introduction of a one-million-token context window, currently available in beta. This allows the model to process and retain significantly more information within a single conversation, making it better suited for long documents, large projects, and extended agent-style workflows.
Not just for developers
Beyond programming, Anthropic says Claude Opus 4.6 can be used for everyday work tasks such as analysing data, running research workflows, and creating documents, spreadsheets, and presentations. Within Anthropic’s Cowork environment, the model can handle multiple tasks autonomously, switching between different types of work without requiring constant user input.
The company has also expanded Claude’s compatibility with office tools. Claude’s Excel integration has been upgraded to handle more complex and long-running tasks, while a research preview of Claude in PowerPoint has been introduced, allowing users to generate and structure presentations based on existing data.
New controls and longer-running tasks
Anthropic has introduced several updates to the Claude API alongside the new model. As per the company, developers now have more control over how much reasoning the model applies through adjustable effort levels, which affect speed, cost, and depth of thinking. A new adaptive thinking feature is said to allow the model to decide when deeper reasoning is necessary, rather than applying it uniformly.
To support longer workflows, Anthropic has also added context compaction, a feature that summarises older parts of a conversation to prevent sessions from hitting hard limits.
Focus on safety and reliability
According to Anthropic, Claude Opus 4.6 underwent its most comprehensive safety evaluation so far. The company says the model shows low rates of misaligned behaviour and fewer unnecessary refusals when responding to safe queries.
Additional safeguards have also been introduced in areas such as cybersecurity, where the model’s stronger capabilities could otherwise be misused.
Availability
Claude Opus 4.6 is available on Claude’s website, through Anthropic’s API, and on major cloud platforms.