EXPLORINEER
en

Archive — US — techcrunch

Anthropic launches Claude Sonnet 4.5, its best AI model for coding

Anthropic has launched Claude Sonnet 4.5, its latest AI model for coding, which promises state-of-the-art performance on coding tasks with capabilities for building production-ready applications. Available via API and maintaining the same pricing as its predecessor, this model aims to compete against OpenAI's GPT-5 amidst a fast-paced AI development landscape.

Key points

  • Anthropic launched Claude Sonnet 4.5 on October 24, 2023.
  • It boasts improved reliability and the ability to build production-ready applications.
  • Pricing remains $3 per million input tokens and $15 per million output tokens.
  • Claude 4.5 is reported to outperform previous models and is already favored by companies like Apple and Meta.
  • Claude Sonnet 4.5 can autonomously build applications over long periods and manage comprehensive tasks.
  • Improved alignment and reduced sycophancy and deception compared to earlier models.
  • Launch of the Claude Agent SDK and a temporary research preview, 'Imagine with Claude'.
  • Anthropic's rapid development cycle reflects increasing competition in the AI space, particularly against OpenAI's GPT-5.

Why it matters

  • This launch signifies a major trend in AI development focusing on generative coding models, which is on the rise as more companies implement AI in their development processes.
  • Reflects a broader industry move towards enhancing AI models' reliability and real-world applicability, a crucial 2025 IT trend.
  • Competition between AI companies like Anthropic and OpenAI resembles past tech rivalries such as those between Apple and Microsoft, affecting market dynamics.
  • The quick turnover of models also highlights the escalating arms race for superior AI capabilities, keeping pace with changing technological demands.

Claude Sonnet 4.5 demonstrates Anthropic's commitment to leading the AI coding frontier, showcasing significant enhancements that may redefine industry standards, especially in the intense competition with models like OpenAI's GPT-5.