On Monday, Anthropic introduced a new AI model named Claude Sonnet 4.5. This model aims to deliver top-notch performance for coding tasks. Unlike its predecessors, it’s designed to create complete applications that are ready for use, rather than just early drafts.
Claude Sonnet 4.5 will be accessible through the Claude API and in the Claude chatbot. Pricing remains stable at $3 per million input tokens—equivalent to about 750,000 words. Output tokens are priced at $15 per million.
Over the past year, developers and enterprises have increasingly favored Anthropic’s models. Big names like Apple and Meta are reportedly using Claude AI models internally. The company has built a strong business by providing API access for coding tools like Cursor and Replit. Recently, competition has grown with OpenAI’s GPT-5, which has exceeded Claude’s models on certain coding tests.
Anthropic claims that Claude Sonnet 4.5 scores highest on various coding benchmarks, such as those verified by SWE-Bench. However, David Hershey, an AI researcher at Anthropic, notes that performance might be challenging to quantify solely through benchmarks. He has observed Claude Sonnet 4.5 working autonomously for up to 30 hours, successfully building an app and even purchasing domain names.
Experts in the field have shared their perspectives on this launch. Micheal Truell, CEO of Cursor, highlighted that Claude Sonnet 4.5 excels particularly in long-term coding tasks. Similarly, Windsurf’s CEO, Jeff Wang, called it a “new generation of coding models.” These endorsements underscore the model’s potential impact in software development.
Anthropic asserts that this version of Claude is more reliable and less prone to generating misleading information than earlier versions. The company believes enhancements have made it more resilient against certain types of attacks.
In conjunction with this launch, Anthropic also unveiled the Claude Agent SDK, a tool for developers to create their own AI agents. Additionally, a new feature called “Imagine with Claude” offers a unique experience for Max subscribers by allowing the AI to generate software in real-time, responding to user prompts without prewritten scripts.
As AI technology evolves rapidly, many companies are racing to release new models frequently. Claude Sonnet 4.5 arrives just two months after the previous model, Claude Opus 4.1. This fast pace makes it tricky for any single company to maintain a leading edge in the market.
While Claude Sonnet 4.5 shows promise, continued advancements from competitors will shape the future of AI coding models. As this landscape evolves, developers and companies will need to stay updated and adapt to these changes.
For more details on coding benchmarks, you can visit SWE-Bench and explore the latest insights from trusted sources like TechCrunch.
Source link
ai coding,AI model,Anthropic,Claude,vibe coding
