Anthropic Unveils Claude Opus 4.5 with Major AI Upgrades

Washington, November 25 (QNA) - Anthropic, the US-based artificial intelligence company, has announced the launch of Claude Opus 4.5, its latest and most advanced large language model to date, featuring substantial improvements in coding performance, agent capabilities, deep research, and handling productivity tools such as spreadsheets and presentation slides.

The company described the new model as "the best model in the world for coding, agents, and computer use," noting that it delivers meaningful gains over previous versions while using significantly fewer tokens. According to Anthropic, Opus 4.5 can use up to 76% fewer output tokens than earlier models while still outperforming them.

Anthropic said the release introduces enhanced developer tools within the Claude Code framework and upgrades to user-facing Claude applications. These improvements aim to support more capable autonomous agents and expand practical integrations across Excel, Chrome, and desktop environments, allowing the model to assist directly within familiar software workflows.

The company emphasized its focus on security and robustness, highlighting significant progress in mitigating malicious use, particularly prompt-injection and command-injection attacks, which embed harmful instructions into websites or data sources to bypass safeguards. Anthropic said Opus 4.5 is its "most robustly aligned model to date," though it acknowledged that no system is completely immune to sophisticated attacks. New internal and external testing frameworks have been added to better measure resilience across coding, browser, and desktop scenarios.

Claude Opus 4.5 is now available through Anthropic's applications, APIs, and major cloud platforms, including Google Cloud's Vertex AI, Amazon Bedrock, and Microsoft's AI services.

 

Source: QNA