Google's newest Flash-class model, Gemini 3.8 Flash, has been added to the model lineup in GitHub Copilot. It sits in the fast, lightweight tier, yet GitHub's early testing highlighted its strength on complex work driven through the terminal. Here is what plans it covers, where it runs, and how billing is handled.

How it performed on terminal-driven work

The framing GitHub chose is worth noting. Rather than leading with aggregate benchmark scores, the announcement points to behavior on complex terminal-based coding tasks.

Two qualities are called out. The first is rigorous validation. The second is persistent recovery from actionable failures. In work that loops between running a command, reading the result, and deciding the next move, what matters is less the quality of any single generation and more whether the model notices a failure and corrects course on its own. Flash-class models are usually sold on speed and cost, so presenting one as durable enough for that grind is a shift from how lightweight models have typically been positioned.

Gemini 3.7 Flash, the previous generation, was added to Copilot in mid-August, so the gap between Google's fast model releases and their arrival in Copilot has grown quite short.

Plans and supported environments

Gemini 3.8 Flash is available to Copilot Pro, Pro+, Max, Business, and Enterprise users. Free tiers are not included.

You select it from the model picker in the following environments.

  • Visual Studio Code
  • Visual Studio
  • Copilot CLI
  • GitHub Copilot cloud agent
  • GitHub Copilot app
  • JetBrains IDEs
  • Xcode
  • Eclipse

Copilot CLI and the cloud agent being in scope from the start lines up with the terminal-focused evaluation above. Rollout is gradual, however, so the model may not appear in your picker immediately. If you do not see it, wait a while and check again.

Billing and administrative control

Billing runs through usage-based pricing, with introductory provider pricing applied through December 31, 2026. Because the price is tied to a fixed window, it is worth keeping in mind that terms could change after the new year.

For organizations, Copilot Enterprise and Copilot Business plan administrators can manage access through the model policy in Copilot settings. The default behavior deserves attention: new models are enabled automatically. The exception is when an administrator has turned off the global default or explicitly disabled the model. In an organization that has never touched these settings, the option simply appears without anyone acting. Teams that want to keep the model list narrow will need a habit of reviewing settings each time something new lands.

Summary

Gemini 3.8 Flash is now usable in GitHub Copilot. It covers the paid plans from Pro through Enterprise and reaches the major IDEs along with the CLI and the cloud agent. Introductory pricing applies through December 31, 2026, and organizations should be aware of the automatic enablement default. Given that it is pitched as a model that holds up under repetitive terminal work, it is worth a try when speed is the priority.

* The thumbnail image is an AI-generated illustration.