Anthropic has announced the release of Claude Sonnet 4.5, which is now available through the Claude API and in Claude Code and other Claude apps. This model is presented as an advancement in coding, complex agent construction, and general computer interaction. The release also includes upgrades to Anthropic's product suite, such as checkpoints in Claude Code, a refreshed terminal interface, and a native VS Code extension. The Claude API has been updated with context editing and a memory tool, while Claude apps now integrate code execution and file creation directly into conversations. The Claude for Chrome extension is also available to Max users who were on the waitlist.
Key Points
- Claude Sonnet 4.5 is available immediately, with pricing remaining consistent with Claude Sonnet 4 at $3 per million input tokens and $15 per million output tokens.
- The model achieved a 61.4% score on the OSWorld benchmark for real-world computer tasks, an increase from Sonnet 4's 42.2% four months prior.
- Claude Sonnet 4.5 is described as state-of-the-art on the SWE-bench Verified evaluation, which assesses real-world software coding abilities.
- The model has demonstrated the ability to maintain focus for over 30 hours on complex, multi-step tasks.
- Anthropic is providing a Claude Agent SDK, offering developers the infrastructure used to power Claude Code.
- Early testing by customers indicated that Claude Sonnet 4.5 reduced average vulnerability intake time for Hai security agents by 44% and improved accuracy by 25%.
- For Devin, Claude Sonnet 4.5 increased planning performance by 18% and end-to-end evaluation scores by 12%.
Context
According to Anthropic, Claude Sonnet 4.5 represents a significant improvement in several areas, including reasoning and mathematics. The model shows enhanced domain-specific knowledge and reasoning compared to older models, including Opus 4.1, as observed by experts in finance, law, medicine, and STEM. The company states that this is the most aligned frontier model it has released, with large improvements in alignment compared to previous Claude models.
Why It Matters
This release indicates a focus on enhancing AI capabilities for software development, agentic workflows, and general computer interaction. Builders can leverage the improved performance in coding and complex task execution, potentially reducing development time and increasing efficiency in various applications.
What To Do
- Open the Claude API documentation to explore the new context editing and memory tool features.
- Test claude-sonnet-4-5 via the Claude API for coding and agentic tasks.
- Note the pricing structure for Claude Sonnet 4.5 and compare it with previous Claude models for cost-effectiveness.
- Watch the provided demo of Claude working in a browser, navigating sites, and filling spreadsheets to understand its upgraded capabilities.
Keep Exploring
/atlas/claude-family