Anthropic Rolls Out Opus 4.5 with Google Chrome and Excel Tools

Anthropic is ending its 4.5 model rollout with the Opus 4.5, which it calls its most advanced version of Claude yet. If Sonnet 4.5 and Haiku 4.5 were the teasers, that means Opus is the blockbuster. It just became the first AI model to score above 80% on SWE-Bench, a benchmark considered one of the hardest tests of real-world coding performance.

That puts Claude nose-to-nose with OpenAI’s GPT-5.1 and Google’s Gemini 3, both of which arrived this month. But Anthropic isn’t trying to win the hype war with big model claims; instead, it's showing off how the model fits into everyday work.

Alongside the model update, Anthropic is launching Claude for Google Chrome and Claude for Excel, moving the assistant directly into the tools people actually use. The Chrome extension will be available to all Max users, and the Excel tool will roll out across Max, Teams, and Enterprise.

What this means is that you'd be able to summarize articles without leaving the browser, automate spreadsheets, formulas, and reports right in Excel, and research, draft, and clean data as you work

While GPT and Gemini are busy flexing multi-modal tricks, Anthropic seems to be saying: AI should sit next to your cursor, not just inside a chat box. The model also introduces an “endless chat” feature, meaning conversations no longer cut off when the AI runs out of space to remember. Instead, it compresses memory on the fly, without interrupting users.

It’s the kind of upgrade that feels small until you realize how many tasks it solves: ongoing software debugging, long research threads, multi-week writing projects, complex legal reviews, and multi-document strategy work.

Dianne Na Penn, Anthropic’s head of product for research, summed it up nicely: “Knowing the right details to remember is really important in complement to just having a longer context window.” So Claude isn’t just remembering more, it’s remembering better.

Opus 4.5 is also designed to act as a lead agent, orchestrating smaller Claude models (like Haiku) to complete bigger tasks. Instead of one big AI doing everything, we’re moving toward AI teams, where Opus plans, decides, and delegates, and other “sub-agent” models do the execution

It’s a direction that could redefine how automation is built, especially for enterprise systems, coding tools, and research platforms.

The takeaway

Opus 4.5 doesn't seem like just another milestone in the AI model arms race. With Excel and Chrome integrations, memory compression, and agent-style performance, Anthropic is positioning Claude as a productivity engine, not just a smart chatbot. In a market where AI companies compete to look impressive, Anthropic is quietly betting that the future belongs to the models that help people finish the report, ship the code, close the analysis, and get the work done.

Anthropic Confirms Claude Was Used in a Major Semi-Autonomous Cyberattack

The attackers’ strategy was a multi-phase misdirection, engineered to jailbreak the model’s safety guardrails.

TechloyOgbonda Chivumnovu