Anthropic introduced Claude Opus 4 and Claude Sonnet 4, its latest generation of hybrid AI models optimized for coding tasks and solving intricate problems.
Claude Opus 4 is so far the most powerful model AI Anthropik, according to the company’s announcement and capable of constantly working on long -term tasks for “several hours”. Anthropic said in tests of customers that Opus 4 was autonomous for seven hours, significantly expanding the capabilities of AI agents. The company also described its modern flagship schedule “The best coding model in the world”, with Anthropic, showing that Opus 4 exceeded Google’s Gemini 2.5 Pro, O3 OpenAI reasoning and GPT-4.1 models in coding and using “tools” such as network search.
Claude Sonnet 4 is a more affordable model focused on performance, which is better suitable for general tasks, which replaces the 3.7 Sonnet model released in February. Anthropic says that Sonnet 4 provides “excellent coding and reasoning”, while providing more true answers. The company adds that both models are 65 percent less often take the abbreviation and gaps to perform tasks compared to 3.7 sonnet and are better in storing key information for long -term tasks, when programmers provide Claude with access to local access to files.
The modern function introduced for both Claude 4 models is “summary of thinking”, which condenses the process of reasoning of the chatbots in an easily understandable way. The “extended thinking” function is also launched in a beta version, which allows users to switch models between reasoning modes or using tools to improve performance and accuracy of answers.
Claude Opus 4 and Sonnet 4 are available at Antropic API, Amazon Bedrock and the Vertex AI Google Cloud platform, and both models are included in the paid Claude plans with the function of extended beta thinking. For now, free users can access Claude Sonnet 4.
In addition to modern models, the Anthropic Code code Agentic Line tool is now generally available after a circumscribed preview in February. Anthropic also claims that it is changing to provide “more frequent model updates” because the company is trying to keep up with the competition from OpenAI, Google and Meta.