Claude Sonnet 4.5
7 hours ago
- #AI
- #Innovation
- #Coding
- Claude Sonnet 4.5 is introduced as the best coding model in the world, excelling in complex agents, computer use, reasoning, and math.
- New features include checkpoints in Claude Code, a refreshed terminal interface, a native VS Code extension, context editing, and memory tools in the Claude API.
- Code execution and file creation (spreadsheets, slides, documents) are now directly integrated into Claude apps.
- The Claude Agent SDK is released, providing developers with the infrastructure to build their own agents.
- Claude Sonnet 4.5 shows state-of-the-art performance on benchmarks like SWE-bench Verified (77.2%) and OSWorld (61.4%).
- Substantial improvements in domain-specific knowledge and reasoning are noted in fields like finance, law, medicine, and STEM.
- Early customers report significant improvements in coding performance, vulnerability intake time, and litigation tasks.
- Claude Sonnet 4.5 is the most aligned frontier model yet, with improved safety and reduced concerning behaviors.
- Released under AI Safety Level 3 (ASL-3) protections, with safeguards against CBRN risks.
- A temporary research preview, 'Imagine with Claude,' showcases real-time software generation.
- Pricing remains the same as Claude Sonnet 4 at $3/$15 per million tokens.