On Monday, Anthropic launched a new border model Called Claude Sonnet 4.5which claims to provide modern performance in coding points. The company says Claude Sonnet 4.5 is able to create “ready -made” applications, and not just prototypes, which represent a jump in reliability from previous AI models.
Claude Sonnet 4.5 will be available through API Claude and Claude Chatbot. Pricing for developers is the same as Claude Sonnet 4: $ 3 per million input brands (about 750,000 words, or more than the entire Lord of the Rings Series) and $ 15 per million production brands.
Last year, ANTHROPIC AI models have emerged as favorite between developers and businesses, largely due to their strong performance in engineering duties. Apple and Meta use Claude AI models internally and Anthropic has made an important API business sale access to AI encoding applications such as the runner, Windsurf and Replit. Recently, Openai’s GPT-5 challenged Anthropic’s dominance in the area, surpassing Claude models in a variety of coding indicators.
Anthropic says Claude Sonnet 4.5 offers top performance at various coding points including Verified the bench. However, AI researcher David Hershey tells Techcrunch that it is difficult to capture Claude Sonnet’s performance 4.5 for benchmarks only.
Hershey says he saw Claude Sonnet Code 4.5 for up to 30 hours during early tests with some business customers. At that time, he watched the AI model not only to create an application, but to support database services, buy domain names and perform a SoC 2 check to make sure the product was safe.
In a statement shared with TechCrunch Managing Director Micheal Truell, he said Claude Sonnet 4.5 represents the state -of -the -art coding performance, especially in longer horizon work. Windsurf CEO Jeff Wang said in a statement that Claude Sonnet 4.5 represents a “new generation of coding models”.
Anthropic also claims that Claude Sonnet 4.5 is Frontier’s most aligned AI model, with lower slander and cheating rates than previous models. The company says it has also improved Claude’s sensitivity to cause injection attacks.
TechCrunch event
Francisco
|
27-29 October 2025
Along with the launch of Claude Sonnet 4.5, Anthropic also starts Agent Claude SDK. The company says this is the same infrastructure that authorizes the Claude Code and can be used to help developers build their own agents.
Anthropic also releases a temporary preview of the research called “Imagine With Claude” for Max subscribers, which shows the software that creates AI in flight. The company says the model will respond to real -time users’ requests without predetermined functionality or preliminary code.
The intense competition in the AI world has made it public for companies to carry flagship models every few months. Claude Sonnet 4.5 starts less than two months after Anthroping’s latest AI model, Claude Opus 4.1. These rapid production cycles make it difficult for each company to have a meaningful lead for a long time.
