Anthropic has released Claude Sonnet 4.5, describing it as the best coding model in the world. Anthropic says this is the strongest model for building complex agents, the best model at using computers, and it shows substantial gains in reasoning and math.

Anthropic released the new version alongside major upgrades to other products, including Claude Code and a new native VS Code extension. 

Claude Sonnet 4.5 is the latest version of a collection of large language models (LLMs) developed by Anthropic. A major strength of the models is the ability to carry out hybrid reasoning, so users can choose between a quick answer for simple tasks and slower but deeper answers for more complex problems.

When Claude Sonnet 3.5 was released, Anthropic described it as setting industry benchmarks for graduate-level reasoning (GPQA), undergraduate-level knowledge (MMLU), and coding proficiency (HumanEval). This latest version builds on that, and Anthropic describes it as the most aligned frontier model the company has ever released, showing large improvements across several areas of alignment compared to previous Claude models. 

Frontier models are general-purpose AI models that go beyond current models, and can be used for reasoning, creative writing, or as in this case, code generation. Anthropic says Claude Sonnet 4.5 is state-of-the-art on the SWE-bench Verified evaluation, which measures real-world software coding abilities.

SWE-bench is a benchmark for evaluating large language models on real world software issues collected from GitHub. Given a codebase and an issue, a language model is tasked with generating a patch that resolves the described problem.

Anthropic says:

"Practically speaking, we’ve observed it maintaining focus for more than 30 hours on complex, multi-step tasks."

For developers interested in building agents, Sonnet 4.5 is available on the Claude Developer Platform natively, and in Amazon Bedrock and Google Cloud's Vertex AI. Developers can also use Sonnet 4.5 to handle complex coding tasks with Anthropic's coding agent, Claude Code, Anthropic's agentic
coding assistant.

Anthropic says that Claude Sonnet 4.5 also improves computer use. On OSWorld, a benchmark that tests AI models on real-world computer tasks, Sonnet 4.5 now leads at 61.4%, and model also shows improved capabilities on a broad range of evaluations including reasoning and math.

Alongside Claude Sonnet, Anthropic also announced the Claude Agent SDK. This is built on the same infrastructure that powers Claude Code, and can be used to build your own agents. Claude Agent SDK is a collection of tools that until now was called Claude Code SDK. The SDK has been updated to add broader capabilities for building AI agents beyond just coding tasks.

Anthropic has also released Claude Code for VS Code on the Visual Studio Marketplace. This lets developers use Claude directly in the terminal, either by subscription or pay-as-you-go pricing. The extension enables Claude to autonomously explore a codebase, read and write code, and runs Terminal commands with your permission.

The extension includes agentic features like subagents, custom slash commands, and MCP. 

Claude Sonnet 4.5 is available now.  

More Information

SWE Bench On GitHub

Claude Sonnet Webpage

Related Articles

Anthropic Launches Prompt Caching With Claude

Power Up Your CLI With Claude Code

JetBrains Adds Claude Support To AI Assistant

Claude Engineer Amplifies Your Code

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info