Copilot Chat now has a 64k context window with OpenAI GPT-4o

Copilot Chat on GitHub.com, GitHub Mobile, the GitHub CLI, as well as officially supported IDEs now have a 64k token window available when working with OpenAI GPT-4o. With this change, customers working with large files and repositories should expect improved responses from Copilot. This change helps Copilot retrieve more information when executing skills to provide contextually relevant responses.

There is no action required on your part to benefit from this upgrade, it is automatically available for all GitHub Copilot users. For more information, check out our documentation and join the discussion within the GitHub Copilot Community.

Hungry for more? – 128k token window for VS Code Insiders

If you’re using GitHub Copilot with Visual Studio Code Insiders, you have access to an even larger 128k context window – the maximum supported by OpenAI GPT-4o. Download the Insiders build to try it out.

Actions metrics improvements: adding runner labels

Starting today, you can now view runner labels in the Jobs tab of your Actions metrics. You can filter by the runner label to view runner specific metrics and answer questions such as:
– “What is the average queue time for my runner?”
– “Which repositories are using my runner?”
– “Which jobs are using the ubuntu-latest label?”

Performance metrics screen with runner label filter applied

To access the feature, on your organization home page, select Insights near the top of the page, and then select ‘Actions Performance Metrics’ on the left side of the page.
To learn more about GitHub Actions Metrics, check out our public documentation or head to our community discussion to ask questions and provide feedback.

See more

GitHub Models now supports the ability to retrieve structured JSON responses in the UI

GitHub Models now supports the ability to retrieve structured JSON responses from models, making it easier to integrate AI outputs into applications and workflows.

While this functionality was already available via our API, this update adds it to the UI.

JSON Response in GitHub Models Playground

Supported models include OpenAI (except for o1-mini and o1-preview) and Mistral models.

To learn more about GitHub Models, check out the docs. You can also join our dedicated community discussion to discuss this update, swap tips, and share feedback.

See more
View all changes