Anthropic unveils Claude Opus 4.5 with long context memory and Chrome/Excel integration

Anthropic has released Claude Opus 4.5 as the final model in its 4.5 series, following Sonnet 4.5 and Haiku 4.5. It delivers state of the art results across benchmarks like SWE Bench, Terminal Bench, tau2 Bench, MCP Atlas, ARC AGI 2, and GPQA Diamond, and is the first model to pass 80 percent on SWE Bench verified.

Opus 4.5 brings long context memory improvements that make it more reliable with large documents, code bases, and multi step workflows. These changes support the new endless chat feature and help the model manage information more effectively during extended tasks.

Anthropic also updated its Google Chrome and Microsoft Excel integrations. Claude for Chrome is now available to all Max users with infinite chat, while Claude for Excel is available to Max, Team, and Enterprise users with support for pivot tables, charts, and file uploads. The model strengthens agentic workflows, coordinating Haiku powered sub agents and refining its own processes, and offers improved resistance to prompt injection attacks. It is available through Anthropic’s apps and API starting at five dollars per one million tokens.