Anthropic has officially released Claude Opus 4.7, the latest iteration of its flagship AI model designed to tackle the most complex tasks in software engineering and enterprise workflows.
Arriving just two months after the launch of version 4.6, this update focuses on refining how the model thinks and interacts with the physical world through vision. The improvements to its reasoning depth and memory make it a more reliable partner for those looking to automate long-running, multi-step projects.
One of the most significant additions in Claude Opus 4.7 is the “xhigh” effort level. In previous versions, developers often had to choose between speed and deep thinking. This new setting sits between the existing high” and max options, allowing for a better balance between how long the model spends reasoning and the speed of its response.
To help manage the costs associated with these deeper thinking sessions, Anthropic has also introduced task budgets in public beta. This feature allows users to set a token ceiling for a specific job. If a task becomes too complex and threatens to run away with costs, the system will stop before it exceeds the limit.
The vision capabilities of the model have received a substantial upgrade. Claude Opus 4.7 can now process high-resolution images up to 2,576 pixels on their longest edge. This is roughly three times the resolution of previous versions.
Claude Opus 4.7 also introduces better file-based memory, which allows it to maintain context across long sessions. If the AI is writing to a scratchpad or taking notes in a file during a multi-hour coding session, it is now much less likely to lose track of what it was doing.
The model also adopts a more direct and assertive tone. While version 4.6 was known for its warmer, more conversational style, Opus 4.7 is more opinionated and literal. It follows instructions with higher precision, though Anthropic warns that prompts written for older models might need a slight adjustment.
Want to see more of our stories on Google?
P.S. Want to keep this site truly independent? Support us by buying us a beer, treating us to a coffee, or shopping through Amazon here. Links in this post are affiliate links, so we earn a tiny commission at no charge to you. Thanks for supporting independent Canadian media!
