05/2025

You can now define a required JSON schema directly in the prompt editor UI, just like in the Models playground. This helps ensure model responses follow a consistent structure, reducing…

Today’s engineering teams struggle with fragmented knowledge. Having critical context scattered across code, documentation, and inside of team members’ heads can make it hard to get up to speed in…

The C++ language has several levels of context outside of the active file. We have been working to improve the quality of C++ GitHub Copilot completions and have new updates…

CodeQL scans on pull requests for JavaScript, TypeScript, Java, Ruby, and Python are now up to 20% faster. This is powered by our new incremental analysis, which only analyzes new…

Starting May 30, 2025, CodeQL no longer generates code scanning alerts for hardcoded secrets. If you want to detect hardcoded secrets in your repositories, we recommend using secret scanning, which…

Enhancements to Copilot code reviews: personal settings, improved comment quality, and expanded language support.

We’re addressing gaps in the GitHub Copilot billing process to ensure consistency with our existing policy. Previously, some Copilot seat assignments were not billed through the end of the billing…

GPT-4o Copilot has been updated with a new model version that provides more useful and timely suggestions.

Dependabot is now generally available for execution on self-hosted GitHub Actions runners managed within Kubernetes clusters using the Actions Runner Controller (ARC). This setup provides auto-scaling, workload isolation, and improved…

Projects that use Gradle need to include dependencies that are resolved at build time in order to get a full, transitive dependency tree. To make this easier, dependency auto-submission now…

  Following the initial sunset notice, we’ve deprecated the automatic watching of repositories and teams. We made this change in order to: Reduce notification noise: You’ll receive fewer unexpected notifications,…

Anthropic’s latest models, Claude Sonnet 4 and Claude Opus 4, are now available in GitHub Copilot. Claude Opus 4 is Anthropic’s most powerful model yet, excelling at complex problem solving…

On June 6, 2025, we’ll be deprecating the ability to convert issues to discussions based on labels. While this bulk migration option will be removed, you can still manually convert…

The latest AI model from xAI, Grok 3 Mini, is now available on GitHub Models. Grok 3 Mini is a compact reasoning model designed for logic-based tasks and context-aware problem-solving.…

CodeQL is the static analysis engine behind GitHub code scanning, which finds and remediates security issues in your code. We’ve recently released version 2.21.3 of CodeQL. Here’s what’s new and…