Anthropic introduces Claude Sonnet 4.5. This AI model is specifically developed for programming. It excels at reasoning and handling mathematics in that context, outperforming previous LLMs. Anthropic states that Claude Sonnet 4.5 is the most advanced LLM of its kind.
Anthropic launches Claude Sonnet 4.5. This Large Language Model (LLM) from the Claude family is specifically developed for programming. It can independently build complex agents and use computers. Anthropic claims that Sonnet 4.5 is currently the best model of its kind.
Long Focus
Anthropic states that Claude Sonnet 4.5 can maintain its focus for over 30 hours during a real-world test. During that period, the model completed a complex multi-step project and built a communication tool like Slack or Teams. The LLM generated 11,000 lines of code and only stopped when the tool was finished.
Claude Sonnet 4.5 succeeds version 4. The previous version could maintain focus for seven hours straight in May, which was already a great success. With version 4.5, Anthropic takes a big step forward. This allows the model to increasingly assist programmers in extensive and complex projects.
“Better than the Rest”
The American AI specialist compares its model with alternatives such as Opus 4.1 and GPT-5 Codex. These internal tests show that Claude Sonnet 4.5 consistently performs best. Based on this, Anthropic claims to have launched the best LLM of its kind on the market. Among other things, this specialization has compelled Microsoft to embrace Claude within Copilot.
Anthropic integrates Claude Sonnet 4.5 into various tools, including Claude Code. Claude Code also gains several new features, such as checkpoints. This makes it possible to save progress. A native VS Code extension is also coming.
Claude Sonnet 4.5 is the direct successor to version 4 and comes with the same price tag. The new LLM is available immediately via the Claude API, the app, and the new Chrome extension.