Anthropic Unveils Claude Sonnet 4.5: A Game-Changer in Software Engineering
Anthropic has unveiled Claude AI Sonnet 4.5, a significant leap in end-to-end software engineering and practical computer use. The model has demonstrated remarkable improvements in coding tasks, with teams observing over 30 hours of uninterrupted focus.
Claude AI Sonnet 4.5 introduces innovative features like 'Checkpoints' for saving and reverting progress, an updated terminal interface, and a native VS Code plugin. These enhancements allow agents to handle longer contexts and manage more complex tasks autonomously, up to 30 hours continuously.
The model showcases substantial gains across common reasoning and math evaluations. On the OSWorld-Verified dataset, it leads at 61.4%, reflecting stronger tool control and UI manipulation for browser/desktop tasks. With a 1M-context setting, it reached 78.2% accuracy on the SWE-bench Verified dataset. Under a higher-compute setting with parallel sampling, accuracy rose to 82.0%.
Claude AI Sonnet 4.5 also exhibits significant improvements in code quality, security vulnerability detection, and domain-specific knowledge (finance, law, medicine, STEM). Its safety posture is ASL-3 with strengthened defenses against prompt-injection.
Claude AI Sonnet 4.5 is now available on Anthropic API & apps, AWS Bedrock, Google Cloud Vertex AI, and GitHub Copilot. With its enhanced capabilities and real-world performance, it is poised to become the global standard in code models, setting a new benchmark for software engineering and practical computer use.
Read also:
- Web3 social arcade extends Pixelverse's tap-to-earn feature beyond Telegram to Base and Farcaster platforms.
- Over 5,600 Road Safety Violations Caught in Manchester Trial
- Trump praises the robustness of US-UK relations during his visit with Starmer at Chequers, showcasing the strong bond between the two nations.
- Navigating the Path to Tech Product Success: Expert Insights from Delasport, a Trailblazer in the Tech Industry
