GitHub Copilot: Gemini 3.8 Flash Now Available
GitHub CopilotGitHub Copilot added Gemini 3.8 Flash, Google's newest Flash-tier model, to its model picker across every major Copilot surface. In GitHub's early testing, the model performed strongly on complex, terminal-based coding tasks and showed more rigorous validation and recovery from failed steps than earlier Flash models. It ships at introductory provider pricing through December 31, 2026, and is rolling out gradually to Copilot Pro, Pro+, Max, Business, and Enterprise plans.
Featured Video
A video we selected to help illustrate this changelog
Key Takeaways
- Gemini 3.8 Flash follows Gemini 3.7 Flash by roughly three weeks, and GitHub moved quickly to add it to Copilot's model picker.
- GitHub says early testing showed the model performing well on complex, terminal-based coding tasks, with better validation and recovery from failed steps than prior Flash models.
- The model is available across eight surfaces, including VS Code, Visual Studio, Copilot CLI, the cloud agent, the Copilot app, JetBrains IDEs, Xcode, and Eclipse.
- Pricing is locked at introductory provider rates through December 31, 2026, giving teams a cost-effective evaluation window before standard billing kicks in.
- Independent benchmarks reported by outlets like The Register and VentureBeat show Gemini 3.8 Flash narrowing the gap with frontier models on coding and agentic tasks while staying at Flash-tier pricing.
- Business and Enterprise admins should check their model policy settings, since new models default to on unless an admin has disabled the global default or the specific model.
Sources & Mentions
5 external resources covering this update
GOOGLE IS BACK! (Gemini 3.8 Flash)
YouTube
Google's Gemini 3.8 Flash is built for agents, while its Cyber twin hunts vulnerabilities
VentureBeat
With Gemini 3.8 Flash, Google reminds everyone it's still in the race
The Register
Gemini 3.8 Flash rolling out three weeks after last release
9to5Google
Gemini 3.8 Flash and 3.8 Flash Cyber
Hacker News
A New Flash-Tier Model Joins Copilot
GitHub Copilot now offers Gemini 3.8 Flash, Google's newest addition to its Flash model line, as a selectable option across the Copilot model picker. The model arrives roughly three weeks after Gemini 3.7 Flash, continuing Google's rapid cadence of Flash-tier releases, and GitHub is positioning it as the more capable successor for everyday coding work.
Early Performance Signals
GitHub's early testing found that Gemini 3.8 Flash "performed strongly on complex terminal-based coding tasks" and demonstrated "rigorous validation and persistent recovery from actionable failures," meaning the model is more likely to catch its own mistakes mid-task and correct course rather than compounding an error across a longer agentic run. Independent coverage of the underlying model backs this up: Google's own benchmarks show Gemini 3.8 Flash improving over 3.7 Flash across software engineering, multi-step reasoning, and agentic workflows, with results in some cases approaching pricier frontier models.
Where to Use It
Gemini 3.8 Flash is available from the Copilot model picker in Visual Studio Code, Visual Studio, GitHub Copilot CLI, the GitHub Copilot cloud agent, the GitHub Copilot app, JetBrains IDEs, Xcode, and Eclipse. The rollout is gradual, so GitHub notes that developers who don't see it immediately in their picker should check back over the following days.
Pricing and Access
The model is billed at introductory provider pricing under usage-based billing through December 31, 2026, giving teams a window to evaluate it before standard rates apply. It's available to Copilot Pro, Pro+, Max, Business, and Enterprise subscribers.
Administration
For Business and Enterprise customers, new models like Gemini 3.8 Flash are enabled automatically through the organization's model policy unless an administrator has turned off the global default for new models or explicitly disabled this specific model. Administrators who want to confirm or adjust access can do so from their organization or enterprise Copilot model settings.