Anthropic has launched Claude 3.5 Sonnet on GitHub Copilot, marking a significant expansion in access to AI-powered coding assistance for GitHub's global developer community.
The integration brings Claude 3.5 Sonnet's capabilities directly into Visual Studio Code and GitHub.com, making it available to GitHub's user base of more than 100 million developers. According to benchmark tests, the model demonstrates superior performance in real-world programming scenarios, achieving the highest score in its class – 93.7% – on HumanEval, a benchmark for Python function writing from natural language descriptions.
Claude 3.5 Sonnet has also outperformed all publicly available models on SWE-bench Verified, a benchmark measuring how effectively AI models can solve actual GitHub issues.
The rollout will occur gradually over several weeks, with all GitHub Copilot Chat users and organisations gaining access during the public preview period.
The new integration offers several key features for developers:
- Production-ready code generation from natural language descriptions
- Real-time debugging assistance with inline chat functionality
- Automated test suite generation
- Contextual code explanations through hover functionality in VS Code
The technical implementation leverages Amazon Bedrock's cross-region inference capabilities. This infrastructure choice aims to maintain consistent service delivery across GitHub's global user base.