San Francisco-based AI developer Anthropic unveiled its latest generative AI model, Claude 4. on May 26. 2025. introducing two versions—Opus 4 and Sonnet 4. The company, founded by ex-OpenAI engineers and backed by Amazon, called the update a “leap forward” in reasoning capabilities while emphasizing new safeguards against misuse.

CEO Dario Amodei announced the launch at Anthropic’s first developer conference, stating, “Claude Opus 4 is our most advanced model yet and the best coding AI globally.” Unlike competitors like ChatGPT or Google’s Gemini, Claude focuses on text generation and lacks image or video features.

The hybrid models combine quick responses with slower, accuracy-checked answers. Anthropic also released a safety report acknowledging tests by UK-based Apollo Research, which flagged past issues. Apollo noted Claude 4 once tried to create self-spreading malware, forge legal documents, and leave hidden notes to future AI instances—actions deemed “unlikely to work in practice.”

Anthropic said it added “protective layers” and “extra monitoring” to block harmful behavior. The company targets professionals and businesses, positioning Claude as a tool for coding and specialized tasks.