Claude Opus 4.5 Is Official: Infinite Chat, Self-Improving Agents, And More

After launching Claude Sonnet 4.5 (late September) and Claude Haiku 4.5 (mid-October), Anthropic on Monday unveiled a new frontier AI model, Claude Opus 4.5, which features state-of-the-art (SOTA) performance in various tasks. Anthropic said the new model "sets a new standard across coding, office tasks, and computer use," delivering improvements on existing capabilities and introducing new features. For example, Infinite Chat is a new feature available to paid subscribers that fixes one of the top complaints from Claude users: context window limit errors. Anthropic says that Claude Opus 4.5 now lets the AI leverage memory to maintain context and consistency across files.

Anthropic also noted that Claude Opus 4.5 can produce various types of files, includifcopng documents, spreadsheets, and presentations, with "consistency, professional polish, and domain awareness." The new model also supports agents for repetitive tasks by directly controlling the computer or browser. Anthropic said that Claude Opus 4.5 achieves "a new state-of-the-art" for production code, computer use, and sophisticated agents. The latter include Claude for Chrome, which is available to all Max users.

Also, Claude Opus 4.5 agents can autonomously improve their capabilities. They can reach peak performance after four iterations, compared to rivals, which can't match Claude agents after 10 repetitions. The Claude Opus 4.5 agents can retain insights from the past to apply to future work.

With Claude Opus 4.5, Anthropic is bringing Claude Code on desktop in research preview. The new SOTA model also adds new capabilities on the developer platform. When coding, Claude Opus 4.5 "handles ambiguity the way a senior engineer would, without requiring hand-holding from the user." The AI will apparently be able to reason across multiple systems when given a bug and figure out how to fix the code issue.

When it comes to complex enterprise tasks, Claude Opus 4.5 also achieved SOTA performance. "Claude Opus 4.5 achieved state-of-the-art results for complex enterprise tasks on our benchmarks, outperforming previous models on multi-step reasoning tasks that combine information retrieval, tool use, and deep analysis," Genspark CTO Key Zhu said in a statement.

Anthropic highlighted Claude Opus 4.5 for Excel use, calling it their best model for creating spreadsheets and specific financial tasks like modeling and forecasting. Claude Opus 4.5 will be available in Excel via a sidebar chat that can assist the user when working on new files. "Claude Opus 4.5 sets a new standard for Excel automation and financial modeling," Fundamental Research Labs cofounder Nico Christie said in a statement. "Accuracy on our internal evals improved 20%, efficiency rose 15%, and complex tasks that once seemed out of reach became achievable." That's an interesting use case for Claude Opus 4.5, especially considering that Microsoft has Copilot AI features ready for Excel use. Claude for Excel is now available to Max, Team, and Enterprise users, supporting pivot tables, charts, and file uploads.