Anthropic has officially launched Opus 4.5, the newest iteration of its flagship large language model, marking the completion of its 4.5 series. This release introduces significant performance enhancements and crucial integrations for popular platforms like Google Chrome and Microsoft Excel, making advanced AI capabilities more accessible to users.
Advanced Performance and Benchmarks
Opus 4.5 demonstrates state-of-the-art performance across a variety of demanding benchmarks. It excels in coding challenges, including SWE-Bench and Terminal-bench, tool utilization with tau2-bench and MCP Atlas, and general problem-solving as evidenced by ARC-AGI 2 and GPQA Diamond scores. Notably, Opus 4.5 is the first model to surpass an 80 percent score on the verified SWE-Bench coding benchmark, underscoring its advanced capabilities in software development tasks.
Seamless Integrations for Chrome and Excel
A key focus of this release is enhancing computer use and spreadsheet capabilities. Alongside Opus 4.5, Anthropic is making its Claude for Chrome extension and Claude for Excel product widely available, moving them beyond their pilot phases. The Chrome extension will be accessible to all Claude Max subscribers, while the Excel-focused model will cater to Max, Team, and Enterprise users, streamlining workflows directly within these essential applications.
Memory Enhancements and "Endless Chat"
Opus 4.5 also features substantial memory improvements designed to handle long-context operations more effectively. These advancements required significant architectural changes in how the model manages information over extended interactions. Dianne Na Penn, Anthropic’s head of product management for research, emphasized the importance of these changes:
“There are improvements we made on general long context quality in training with Opus 4.5, but context windows are not going to be sufficient by themselves. Knowing the right details to remember is really important in complement to just having a longer context window.”
These memory enhancements have also enabled a highly anticipated “endless chat” feature for paid Claude users. This allows conversations to continue uninterrupted even when the model reaches its context window limit, as the system intelligently compresses its context memory without alerting the user.
Designed for Agentic AI Applications
Many of the upgrades in Opus 4.5 are tailored for agentic AI use cases, particularly scenarios where Opus acts as a primary agent overseeing a group of Haiku-powered sub-agents. Effectively managing such complex tasks demands robust working memory, where the described memory improvements prove invaluable. Penn further elaborated:
“This is where fundamentals like memory become really important, because Claude needs to be able to explore code bases and large documents, and also know when to backtrack and recheck something.”
Opus 4.5 enters a competitive landscape, facing off against other recently launched frontier models such as OpenAI’s GPT 5.1, released on November 12, and Google’s Gemini 3, which debuted on November 18.
