VS Code 1.117 adds bring-your-own AI model keys and smoother Copilot Chat streaming

Microsoft has released Visual Studio Code 1.117, continuing its rapid update cadence with changes aimed at developers using GitHub Copilot and in-editor chat tools. The update focuses on broader model support, faster-feeling chat output, and a fix for launching Copilot CLI across more terminal setups.

A key addition is Bring Your Own Key support for Copilot Business and Enterprise, allowing organizations to connect external model providers using their own API keys. Microsoft says this enables teams to use models from services such as OpenAI, Google, OpenRouter, Ollama, and others directly within VS Code chat.

More control over model choice

For companies, BYOK can simplify compliance and cost management by keeping billing and usage tied to existing vendor accounts. It can also reduce reliance on Microsoft-hosted capacity, including enabling local model workflows where supported, such as through Ollama.

The release also introduces an experimental approach to incremental chat rendering designed to make responses appear more fluid. Instead of timer-based updates, VS Code can stream content block by block with optional animations as tokens arrive, which may make replies feel quicker during longer outputs.

Terminal and Copilot CLI improvements

VS Code 1.117 additionally improves how the Copilot CLI terminal profile starts from the terminal panel. Microsoft says the profile can now launch even when the default terminal is set to a non-default shell like fish on macOS or Linux, or Git Bash on Windows, addressing cases where the terminal previously failed to start.

As with other VS Code releases, updates typically roll out automatically on Windows and macOS, while Linux users may need to use their distribution’s package or update manager. Microsoft has published full release notes detailing the changes and known issues for the 1.117 build.