MAI-Code-1.1-Flash
MAI-Code-1.1-Flash is a coding model integrated into GitHub Copilot, including local execution on supported devices.
What the release provides
It accepts images such as screenshots and diagrams for coding tasks. The current product page describes native VS Code and Copilot CLI integration. Local inference has no per-inference charge for the local model call.MAI-Code-1.1-Flash official model page
The published local checkpoint
Microsoft's November 2026 engineering report describes a 137.9B-parameter model and a released 53.0 GiB mixed-precision checkpoint. That package is available through Microsoft Foundry and local GitHub Copilot. Weight size excludes the additional context cache and runtime memory.MAI-Code local checkpoint engineering report
How to use it well
Practical guidance. Confirm which requests run locally and which use a cloud service. Review the changed files and observed application behavior after a coding task. A local model call does not establish that every part of the surrounding development product is offline.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Provider product, Downloadable |
| Model identifiers | MAI-Code-1.1-Flash |