Anthropic Launched New AI Models Claude Opus 4 And Sonnet 4

On 22 May 2025, Anthropic released two AI models, which it has named Claude Opus 4 and Claude Sonnet 4. These are the latest in the Claude 4 family and have been created to assist developers when facing difficult coding tasks and for complex assignments. Both models involve smarter thinking and work great for long tasks, for software development, and for creating AI agents. They can also bring in tools like web search and access to local files to provide more complete answers.

Table of Contents

Claude Opus 4

Claude Opus 4 is the most advanced model from Anthropic. It is known as the best coding model in the world. It can work for hours on difficult code editing and bug fixing jobs without losing focus. Claude Opus 4 scored 72.5 percent on SWE-bench Verified and 43.2 percent on Terminal-bench. It performed better than top models from OpenAI and Google.

Claude Sonnet 4

This Model is made for everyday use and is faster and more efficient. Claude Sonnet 4 replaces the earlier Sonnet 3.7 model. It scored 72.7 percent on SWE-bench Verified and is used in GitHub Copilot’s new coding tool. It helps with writing clean code, solving problems, and following user instructions better.

Availability and Price

Both models are accessible on Anthropic API, Amazon Bedrock, and Google Cloud Vertex AI. Claude Opus 4 is the option in paid plans and will run you 15 dollars for input and 75$ for output per one million tokens. Claude Sonnet 4 is included in free or paid plans, at 3 and 15$ per month for input and output models, respectively.