Anthropic on Thursday unveiled its latest artificial intelligence model, the Claude Opus 4, which the start-up said can write code for hours at a time.
The company, which has been backed by Google and Amazon, said the new Claude Opus 4 and the Claude Sonnet 4 models set "new standards for coding, advanced reasoning, and AI agents".
Anthropic first launched the Claude chatbot in 2023. The latest Claude Opus 4, Anthropic said, can run autonomously for seven hours with sustained performance. "Claude Opus 4 is our most powerful model yet and the best coding model in the world," Anthropic said.
The Opus 4 model pushes the limits in coding, research, writing and scientific discovery, the company said. Meanwhile, the Sonnet 4 delivers a stronger performance for everyday uses as an upgrade to the Sonnet 3.7
The Claude Opus 4 and Sonnet 4 are both hybrid models that can deliver quick responses to questions or take time for deeper thinking. Both models can also extract and save "key facts to maintain continuity and build tacit knowledge over time".
Benchmarks from Anthropic showed the Opus 4 outperformed OpenAI's o3 and GPT-4.1, as well as Google's Gemini 2.5 Pro, in coding tasks and tool uses.
OpenAI on Wednesday announced Apple design master Jony Ive would be merging his company io with Sam Altman's company. And on Monday, Microsoft said it will add Elon Musk's Grok to in its cloud.
