On Thursday, Anthropic introduced the latest iteration of its advanced AI model, Opus 4.6, which plays a crucial role in enhancing Claude Code. Following the release of Opus 4.5 last November, this update aims to expand the model's functionality and attract a wider range of users.
A standout feature of Opus 4.6 is the introduction of "agent teams." This innovative approach allows multiple agents to collaborate by dividing larger tasks into smaller, manageable segments.
According to the company, "Instead of one agent tackling tasks in sequence, you can distribute the workload among several agents--each responsible for its part while directly coordinating with others." Scott White, Anthropic's Head of Product, likened this feature to having a skilled human team working together, emphasizing that this segmentation enables faster and more efficient collaboration. Currently, agent teams are available in a research preview for API users and subscribers.
Opus 4.6 also boasts an extended context window, allowing the model to retain a greater volume of information during user sessions. With the capability of handling 1 million tokens of context, it can effectively manage larger code bases and documents, similar to the offerings of the company's Sonnet versions 4 and 4.5.
Additionally, the new version integrates Claude directly into PowerPoint, enhancing accessibility. Previously, users had to transfer files to PowerPoint for editing after creating a deck with Claude. Now, presentations can be developed directly within PowerPoint with Claude's assistance.
White shared with TechCrunch that Opus has transformed from a model focused solely on software development to a versatile tool beneficial for a wider array of knowledge workers. He noted that many users, including product managers, financial analysts, and professionals from diverse fields, have found value in Claude Code due to its remarkable task execution capabilities.