Anthropic: Launches Opus 4.5 Powerful New AI Model

Prime Highlight

Opus 4.5 becomes the first model to score above 80% on the SWE-Bench coding benchmark, showcasing major advancements in complex reasoning and code-fixing capabilities.
Anthropic introduces upgraded product integrations, bringing Claude for Chrome to all Max users and Claude for Excel to Max, Team, and Enterprise plans.

Key Facts

Opus 4.5 features improved long-context memory management, enabling an “endless chat” experience by compressing earlier conversation history automatically.
The model is designed to serve as a lead agent in agentic workflows, coordinating tasks with smaller Haiku-powered agents for handling large code bases and documents.

Background

Anthropic on Monday introduced Opus 4.5, the newest and most powerful model in its 4.5 series. The release follows Sonnet 4.5 in September and Haiku 4.5 in October, completing the company’s major lineup upgrade for the year.

Opus 4.5 brings great improvements across several benchmark tests. It is the first model to score more than 80% on SWE-Bench, a highly respected benchmark that tests how well AI can understand and fix code. The model also showed leading performance in tool use and complex reasoning tests such as ARC-AGI 2 and GPQA Diamond.

To highlight these capabilities, Anthropic is rolling out new product integrations. Claude for Chrome and Claude for Excel, which were earlier available only to a limited group, will now reach more paying users. The Chrome extension will be accessible to all Max users, while the Excel version will support Max, Team and Enterprise plans.

One of the most significant upgrades in Opus 4.5 is its improved memory management. Anthropic says the model can now handle long-context tasks more effectively without losing key details. These upgrades also enable an “endless chat” feature for paid customers. Instead of stopping when the context limit is reached, Opus 4.5 compresses earlier conversation information silently so users can continue chatting without disruption.

Anthropic sees these improvements as essential for agentic use cases, where Opus works as a lead agent while assigning tasks to smaller Haiku-powered agents. Better memory helps the system explore large code bases and documents and return to earlier steps when needed.

With this launch, Opus 4.5 enters a competitive market, going head-to-head with OpenAI’s GPT 5.1 and Google’s Gemini 3, both released earlier this month.