Anthropic has launched its next-generation AI models, Claude Opus 4 and Claude Sonnet 4, marking a significant advancement in artificial intelligence that balances cutting-edge capabilities with responsible deployment.
The Claude 4 models, released on May 22, 2025, introduce a hybrid reasoning system that allows users to choose between near-instant responses and extended thinking for deeper reasoning. Both Opus 4 and Sonnet 4 are "hybrid" models capable of near-instant responses and extended thinking for deeper reasoning. With reasoning mode switched on, the models can take more time to consider possible solutions to a given problem before answering. As the models reason, they'll show a "user-friendly" summary of their thought process.
Anthropic debuted these models amid intense competition between Anthropic, Google, and OpenAI for the "best frontier model" crown. Claude Opus 4, described as a "powerful, large model for complex challenges," can perform thousands of steps over hours of work without losing focus. Anthropic claims Claude Opus 4 is "the world's best coding model," with sustained performance on complex, long-running tasks and agent workflows.
Notably, Anthropic has activated AI Safety Level 3 (ASL-3) for Claude Opus 4 to "limit the risk of Claude being misused specifically for the development or acquisition of chemical, biological, radiological, and nuclear (CBRN) weapons." The Amazon-backed company said it hadn't determined if Opus 4 had crossed the benchmark that would require tighter controls but was taking precautions. This decision followed internal findings that Opus 4 could help users with limited expertise produce hazardous materials. In response, more than 100 safety controls have been implemented, including real-time monitoring, restricted data egress, and a bug bounty program.
The company says Claude Opus 4 and Claude Sonnet 4 are defining a "new standard" for AI agents that "can analyze thousands of data sources, execute long-running tasks, write human-quality content, and perform complex actions." Anthropic, founded by former OpenAI research executives, launched its Claude chatbot in March 2023 and has since been part of the increasingly heated AI arms race. Anthropic stopped investing in chatbots at the end of last year and has instead focused on improving Claude's ability to do complex tasks like research and coding.
Both models include "extended thinking," which allows Claude to switch between two modes: deep reasoning and action performance. Claude can run data analysis as needed, improving accuracy as it works, which helps it better anticipate and execute next steps. Claude models can now use tools like web search while engaging in deeper reasoning, letting them switch between thinking and tool use to generate better results. They can run tools simultaneously, follow instructions with greater accuracy, and show enhanced memory—picking up and storing key information to ensure better context retention.
Anthropic's annualized revenue reached $2 billion in the first quarter of 2025, more than doubling from a $1 billion rate in the prior period. Pricing for Opus 4 is set at $15 per million input tokens and $75 per million output tokens, while Sonnet 4 is priced at $3 for input and $15 for output per million tokens.
As AI capabilities continue to advance, Anthropic's approach demonstrates that powerful models can be developed responsibly, with safety measures that scale alongside increasing capabilities.