Anthropic Releases Opus 4.5, Delivering Breakthrough AI Performance and Enhanced Agentic Capabilities

Anthropic has introduced Opus 4.5, the final and most advanced model in its 4.5 series, following earlier releases of Sonnet 4.5 and Haiku 4.5. The new flagship model sets state-of-the-art results across major AI benchmarks, including SWE-Bench, Terminal-bench, tau2-bench, MCP Atlas, ARC-AGI 2, and GPQA Diamond. Opus 4.5 is also the first model to surpass 80% on SWE-Bench verified, marking a milestone in coding performance.

Alongside the release, Anthropic is broadening access to Claude for Chrome and Claude for Excel, expanding availability to Max, Team, and Enterprise users. The company has also redesigned Opus’s memory system to support stronger long-context reasoning and introduce an “endless chat” feature that allows conversations to continue seamlessly as context compresses automatically.

The upgrade strengthens Opus’s role in agentic workflows, enabling the model to direct Haiku-powered sub-agents and manage complex tasks such as codebase navigation and large-document analysis. Opus 4.5 enters a competitive frontier model landscape, launching shortly after OpenAI’s GPT-5.1 and Google’s Gemini 3.