GitHub's changelog for November 13, 2026 announces the general availability of Gemini 4 Hydrogen inside GitHub Copilot. Google positions the model as the latest member of its Flash line, which prioritizes speed and efficiency over maximum reasoning depth. According to GitHub, it is built for agentic coding, tool calling and multistep workflows, and it follows instructions more reliably than its predecessor, Gemini 3.8 Flash.

Availability and surfaces

The model is offered to subscribers on the Copilot Pro, Pro+, Max, Business and Enterprise plans. Users can choose it from the model picker in a wide range of environments: Visual Studio Code, Visual Studio, the Copilot CLI, the GitHub Copilot coding agent, the GitHub Copilot app, github.com, GitHub Mobile on iOS and Android, JetBrains IDEs, Xcode, Eclipse and the Copilot SDK. That breadth shows how far Copilot has expanded beyond a single editor plugin into a platform spanning IDEs, terminal, web, mobile and programmatic access. GitHub notes the rollout is gradual, so some accounts may not see the model immediately.

Pricing and billing

Gemini 4 Hydrogen is billed at the provider's list pricing under Copilot's usage-based billing. This indicates that GitHub now passes through model-level costs for at least some models instead of relying only on flat premium request multipliers. For teams, the practical implication is that choosing a fast, cheaper Flash-class model for routine agentic tasks can be an effective cost-control lever, while reserving more expensive frontier models for harder problems. GitHub points users to its models and pricing documentation for exact rates.

Administrative controls

For Copilot Business and Enterprise, administrators manage access through the model policy in Copilot settings. Under default model enablement, new models switch on automatically unless the administrator has disabled the global default or has explicitly blocked the specific model. Organizations with strict governance or data-residency requirements should review these settings, since a newly released model can become available to developers without any additional action.

Context within the Copilot model lineup

The related-posts list on the changelog shows a rapid cadence of model additions. Gemini 4 Carbon arrived on November 10, and Claude Fable 5.2 was added on November 9. A separate notice on November 13 warns of upcoming deprecations of selected Copilot models in December. Together these items illustrate GitHub's strategy of operating Copilot as a multi-vendor model hub, frequently adding new releases from Google and Anthropic while retiring older ones. The pairing of Gemini 4 Carbon and Gemini 4 Hydrogen within days of each other suggests Google now has a tiered presence in Copilot, with a heavier model and a lighter, faster Flash option.

Industry implications

Flash-class models matter increasingly in agentic coding because agents make many sequential calls, and latency and cost compound quickly. A model tuned for tool use and instruction following can make autonomous workflows, such as the Copilot coding agent or CLI sessions, more responsive and affordable. By making Gemini 4 Hydrogen generally available rather than in preview, GitHub signals confidence in its stability for production use. For Google, inclusion across every major Copilot surface is a distribution win against competing model vendors in the developer tools market. For rivals such as Anthropic and OpenAI, it underscores that Copilot users can switch models easily, so competition is decided by measurable quality, speed and price rather than by platform lock-in.

The announcement itself contains no benchmark data, so claims about improved instruction following come from the vendor description and have not been independently verified here. Developers will likely judge the model through real-world use in agent mode, where speed and reliable tool invocation are the key differentiators. Teams should evaluate it against their existing default models before changing standard configurations.