What happened
GitHub’s November 9, 2026 changelog says Gemini 4 Carbon, described there as Google’s latest model, is now generally available in GitHub Copilot. GitHub positions the model for complex coding, agentic and reasoning tasks. It also says the model has a focus on secure code generation and security-focused work. The announcement is an availability update: it does not announce a new Carbon version, describe changes to the underlying model, or publish comparative performance results.
GitHub says the option is available to Copilot Pro+, Max, Business and Enterprise users. Eligible users select it through the model picker rather than a separate product or installation. The changelog lists Visual Studio Code, Visual Studio, Copilot CLI, GitHub Copilot coding agent, the GitHub Copilot app, JetBrains IDEs, Xcode and Eclipse as the places where the picker offers Carbon. This puts the same model option across interactive editor workflows, command-line use and GitHub’s coding-agent surface.
Availability is not necessarily simultaneous for every eligible user. GitHub says the rollout will be gradual and tells users to check back if they do not yet see the model. The changelog does not provide a rollout timetable, identify which clients or plans receive it first, or say whether availability differs by region. It also does not state whether a user needs to update a particular client before Carbon appears in the picker.
GitHub says Gemini 4 Carbon is billed at provider list pricing under usage-based billing. It directs customers to Copilot’s models-and-pricing documentation for details, but gives no specific price, unit of billing, included allowance or spending example in the announcement itself. For a team, the practical implication is that access to a new model does not establish its operating cost; the applicable plan and model-pricing information still determine that.
For Copilot Business and Copilot Enterprise, administrators can manage access to Carbon through the model policy in Copilot settings. Under GitHub’s default model-enablement setting, new models are automatically enabled unless an administrator has disabled the global default or explicitly disabled this model. That makes the release both a user-facing picker addition and an administrative policy event. GitHub also points readers to its documentation for the broader set of models available in Copilot and invites feedback through GitHub Community.
What the archive adds
New Runtime’s earlier coverage said Gemini 4 Carbon moved from testing and early access to public availability. This is a new fact on that known story: GitHub’s changelog specifies Carbon’s availability inside Copilot, eligible plans, client surfaces, usage billing and administrator controls.
Trend signal
This supports the pattern that harness architecture outlives model choice. Carbon enters through Copilot’s existing model picker, clients, coding-agent surface, billing arrangement and policy controls; those harness layers determine how a team can use and govern it. The announcement establishes access, not whether Carbon outperforms another model or changes the quality of completed work.
Why now
GitHub dated the general-availability announcement November 9, 2026. Its stated delivery mechanism is a gradual rollout rather than a single moment when every eligible account sees the choice. The changelog gives no additional explanation for the date, the rollout sequence or the decision to add Carbon at this point.
Connections to other events
An earlier New Runtime note from October 26 covered GPT-6.1 Terra becoming generally available in GitHub Copilot across IDEs, CLI and cloud agent. Carbon is another example of Copilot expanding selectable models across those product surfaces, while retaining the platform’s billing and administrative layer.
Why it matters
For teams already using Copilot, Gemini 4 Carbon can be evaluated inside an established development environment instead of requiring a switch to a separate coding tool. Its placement in IDEs, Copilot CLI and the coding agent means an evaluation can cover both interactive development and delegated coding tasks. GitHub’s stated security focus may make security-oriented tasks a reasonable test category, but it is not evidence that the model produces safer code in a particular repository.
The operational details matter as much as model selection. A Business or Enterprise administrator can allow or block the model through policy, while provider-list, usage-based billing means adoption can affect spending. Teams should establish which users can select Carbon, verify the applicable rate, and test it on representative tasks before making it a default. Review effort, latency, completion quality and cost are separate questions that the availability announcement does not answer.
What happens next
- More eligible Copilot accounts will receive the Carbon model picker option as the gradual rollout proceeds. (3 months, 75%). Watch for: GitHub updates to the changelog or Copilot model documentation.
- Business and Enterprise teams will make local enablement decisions through Copilot model policy. (3 months, 65%). Watch for: administrator guidance or policy-control updates from GitHub.
Limits and uncertainty
GitHub’s announcement gives no per-token or per-request prices, benchmark results, latency figures, usage limits or security evaluation for Gemini 4 Carbon. It says the model focuses on secure code generation and security-focused work, but does not identify tests, repositories, threat models or measured outcomes behind that positioning. The rollout is gradual, and GitHub does not say when all eligible accounts and listed clients will receive access. The changelog also does not independently compare Carbon with other Copilot models. Its claims establish product availability and intended use, not superior coding, reasoning or security results.
Skill to build now
Build a small, repeatable evaluation around one task your team handles often. Run the same repository issue or security-focused coding task through Carbon and the model currently used in Copilot, then record whether the task is completed, how many review corrections are needed, elapsed time and billed usage. For a Business or Enterprise team, first confirm the relevant model policy with the administrator. This turns a picker change into a decision based on the team’s own workload.
Reader takeaway
Gemini 4 Carbon is now a Copilot model option for eligible plans across multiple development clients, but its rollout is gradual and its charges follow provider-list, usage-based pricing. Treat the release as an opportunity for controlled comparison on real work. For organizational accounts, model policy, review capacity and cost checks belong in the rollout plan.