GitHub Copilot adds memory and new model choices for devs
Updated on December 29, 2025 4 minutes read
Frequently Asked Questions
What is GitHub Copilot memory?
Copilot memory is a public preview feature that lets Copilot agents learn from a repository over time and reuse that context. It’s designed to improve results across Copilot coding agent and Copilot code review, and it can be enabled from GitHub Settings under Copilot.
Which models can I choose for Copilot coding agent?
For Copilot Pro and Copilot Pro+, GitHub’s model picker lets you choose Auto, Anthropic Claude Opus 4.5, Anthropic Claude Sonnet 4.5, or OpenAI GPT-5.1-Codex-Max. Auto selects a model based on availability, and you can switch models per task.
How does Auto model selection affect premium requests?
In Visual Studio Code, Auto routes requests to available models and bills premium request usage based on the model chosen. GitHub says Auto is currently limited to models with 0× to 1× multipliers, and paid subscribers get a 10% multiplier discount when using Auto.
Do I need to update my IDE to use the newest Copilot models?
Some models have minimum client requirements. GitHub’s model reference notes that GPT-5-Codex requires VS Code v1.104.1+, and certain GPT-5.1 Codex variants require specific plugin versions across VS Code, JetBrains, Xcode, and Eclipse, so keeping IDEs and extensions updated matters.