Anthropic Launches Claude Opus 4.5
Anthropic's newest model, Claude Opus 4.5, is available today. The company calls it "the best model in the world for coding, agents, and computer use."
It's also meaningfully better at everyday tasks like deep research and working with slides and spreadsheets. Anthropic positions Opus 4.5 as a step forward in what AI systems can do, and a preview of larger changes to how work gets done.
Microsoft and Nvidia announced multi-billion-dollar investments in Anthropic last week, boosting the AI lab's valuation to about $350 billion. Claude Opus 4.5 is available on Anthropic's apps, API, and on all three major cloud platforms. Pricing is $5 per million input tokens and $25 per million output tokens, significantly cheaper than the previous Opus at $15/$75.
Claude Opus 4.5 achieves state-of-the-art performance in agentic coding, outperforming rival models like Google's Gemini 3 Pro and OpenAI's GPT-5.1 on SWE-bench Verified, a test set that measures an AI system's software coding abilities. The model scores 80.9% on SWE-bench Verified and reaches 66.3% on OSWorld for computer use tasks.
Anthropic said it tested Claude Opus 4.5 on a difficult take-home exam that it gives to prospective performance engineers, and the model scored higher than any human candidate ever had. Early testers report that Claude Opus 4.5 can complete tasks that were impossible for Sonnet 4.5, handling ambiguity and reasoning about tradeoffs without hand-holding.
Claude Code gains two upgrades with Opus 4.5. Plan Mode now builds more precise plans and executes more thoroughly. Claude Code is also now available in the desktop app, letting users run multiple local and remote sessions in parallel.
For Claude app users, long conversations no longer hit a wall as Claude automatically summarizes earlier context as needed. Claude for Chrome, which lets Claude handle tasks across browser tabs, is now available to all Max users. Beta access to Claude for Excel has expanded to all Max, Team, and Enterprise users.
The core characteristics of Opus 4.5 include a 200,000 token context window, 64,000 token output limit, and a March 2025 knowledge cutoff. The model offers hybrid reasoning that allows for instant responses or extended thinking. API users have fine-grained controls for adjusting the overall effort it puts into a response, balancing performance with latency and cost.
Opus 4.5 is also available in public preview in Microsoft Foundry, GitHub Copilot paid plans, and Microsoft Copilot Studio.