Anthropic Unveils Claude Opus 4.5

Anthropic has officially released Claude Opus 4.5, marking its most capable large language model (LLM) to date with advanced coding performance and enterprise productivity features.

The flagship model achieved 80.9% on SWE-bench Verified, establishing it as the current leader in AI code generation and competing directly with OpenAI’s latest models. Opus 4.5 excels in agentic workflows, demonstrating improved task refinement that reaches peak capability in just four iterations compared to 10 attempts required by competing models.

Pricing for the new model starts at $5 per million input tokens and $25 per million output tokens via the Claude API. The model is available today through Anthropic’s apps, API, and all three major cloud platforms using the identifier claude-opus-4-5-20251101.

Browser and Spreadsheet Integration Expands Productivity

Opus 4.5 brings significant enhancements to real-world computer tasks, building on the Chrome extension capabilities introduced with Anthropic’s Claude Haiku 4.5 in October. The Chrome extension, available to Max plan users, enables Claude to navigate websites, fill forms, and complete multi-step workflows without manual copying between windows.

Excel automation represents a major focus for the new model. In early customer testing, Anthropic reported 20% accuracy improvements and 15% efficiency gains for Excel tasks. The model understands entire workbooks including nested formulas and multi-tab dependencies, providing explanations with cell-level citations.

Claude for Excel, currently in beta through a waitlist limited to 1,000 Max, Team, and Enterprise customers, handles financial modeling, error debugging, and scenario testing while maintaining formula structure. The integration does not yet support pivot tables, conditional formatting, macros, or VBA.

Positioning Within the Claude 4.5 Family

Opus 4.5 completes Anthropic’s model lineup following Sonnet 4.5’s September release and Haiku 4.5’s October debut. Within the family, Opus handles core agentic tasks and production code requiring maximum sophistication, Sonnet powers customer-facing agents at scale with low latency, and Haiku provides an accessible entry point for basic tasks.

The model’s long-running agent capabilities enable extended work periods on complex tasks like application rewrites and deep analysis. Beyond coding, Opus 4.5 delivers broad improvements in complex information retrieval, agentic tool use, and enterprise automation.

Anthropic’s release positions the company among the best AI assistants and strengthens its position in the AI tools for business market. The Chrome integration places Claude alongside other browser-based AI tools, while the Excel features compete directly with AI-powered data analysis tools.

The launch follows a period of rapid advancement for Anthropic. The company announced a strategic partnership with Microsoft and NVIDIA on November 18, enabling Claude models to integrate into enterprise platforms including Microsoft 365 Copilot. This infrastructure expansion supports the broader deployment of Opus 4.5 across corporate environments.

Developers can access the new model immediately through Anthropic’s API, with enterprise customers able to deploy through their existing cloud infrastructure. The company has not announced a specific timeline for expanding the Excel beta beyond its current waitlist.