GitHub Copilot model picker showing MAI-Code-1-Flash selected Image: GitHub / github.blog
by Michael Joiner

MAI-Code-1-Flash Is Now in Copilot CLI, Mobile, and JetBrains; Opus 4.6 Fast Retires June 29

Two model changes hit GitHub Copilot on June 18: Microsoft's MAI-Code-1-Flash expanded from VS Code to eight more surfaces, and GitHub announced Opus 4.6 (fast) will be deprecated across all Copilot experiences on June 29.

Share

GitHub Copilot had two model-related announcements on June 18. MAI-Code-1-Flash, Microsoft’s purpose-built coding model, expanded from VS Code to eight additional surfaces. And GitHub announced that Opus 4.6 (fast) will be deprecated across all Copilot experiences on June 29.

MAI-Code-1-Flash Is Now Everywhere

MAI-Code-1-Flash launched on June 2 in VS Code only. The June 18 update rolls it out to the rest of Copilot’s surfaces:

  • Copilot CLI
  • GitHub Copilot app
  • Copilot Chat on GitHub
  • Visual Studio
  • GitHub Mobile
  • JetBrains IDEs
  • Eclipse
  • Xcode

The model is available on Free, Student, Pro, Pro+, and Max plans. GitHub is rolling it out gradually, starting with a limited user set. Business and Enterprise access is coming, though no date has been set.

What MAI-Code-1-Flash Is

MAI-Code-1-Flash is the first coding model built entirely inside Microsoft, by the MAI division led by Mustafa Suleiman. It’s a sparse Mixture-of-Experts model with 137 billion total parameters, a 256K context window, and API pricing of $0.75 input / $4.50 output per million tokens.

The pitch is efficiency. Microsoft says it handles harder problems with up to 60% fewer tokens than competing small models and outperforms Haiku 4.5 on coding benchmarks. It was designed specifically for how Copilot users actually work, which in practice means frequent, shorter exchanges rather than long-form reasoning sessions.

The model selection in Copilot is gradual by default. Auto mode routes easier completions to models like MAI-Code-1-Flash and reserves heavier models for complex requests. You can also select it explicitly in the model picker if you want to route everything through it.

Opus 4.6 Fast Is Going Away

Also on June 18, GitHub announced the deprecation of Opus 4.6 (fast) across all Copilot surfaces, effective June 29. This covers Copilot Chat, inline edits, ask mode, agent mode, and code completions.

This is not a sudden change. GitHub already removed Opus 4.6 fast from Copilot Pro+ in April, citing usage patterns. The June 29 date cleans up the remaining surfaces that still had it available.

GitHub recommends Opus 4.6 (standard) as the drop-in alternative for anyone who was specifically choosing the fast variant.

Sources: GitHub Changelog — MAI-Code-1-Flash on more surfaces · GitHub Changelog — Opus 4.6 (fast) deprecation · Microsoft AI — Introducing MAI-Code-1-Flash

Share

The Weekly Diff

One email a week: every AI coding tool price change, plan restructure, and major release we verified, with sources. No filler.

Free. Unsubscribe anytime.