GitHub announced the general availability of Gemini 4 Argon in GitHub Copilot on October 13, 2026. The model is positioned for complex coding, agentic workflows, and reasoning-heavy tasks, giving eligible Copilot users another option beyond the models already available in their environment.
The release matters most for teams that use Copilot across different surfaces, including IDEs, the Copilot CLI, coding agent, and Copilot app. Access is not universal by default, however: plan eligibility, client support, organization policy, and the gradual rollout all affect whether the model appears.
What changed in GitHub Copilot
Gemini 4 Argon is now listed as generally available in GitHub’s supported AI model catalog. GitHub describes it as Google’s newest frontier model and says it is designed for complex coding, agentic, and reasoning tasks.
GitHub also reports that early testing showed Gemini 4 Argon solved more tasks than previous Gemini models on every coding benchmark tested, with the largest gains in Copilot CLI. The announcement does not provide the benchmark names, scores, methodology, or scope, so this should be treated as GitHub’s product claim rather than an independently quantified comparison.
GitHub says the model thoroughly checks its work and tests edge cases before finishing. That describes the intended behavior of the product; it is not a guarantee that generated code is correct, complete, or safe to merge without review.
Who can use Gemini 4 Argon
The announcement identifies Copilot Pro+, Max, Business, and Enterprise as eligible plans. It lists the following Copilot surfaces:
- Visual Studio Code
- Visual Studio
- Copilot CLI
- GitHub Copilot coding agent
- GitHub Copilot app
- JetBrains IDEs
- Xcode
- Eclipse
Availability is rolling out gradually, so an eligible user may not see Gemini 4 Argon immediately. The supported-model documentation also includes the model in relevant catalog sections, but the supplied table representation does not preserve every plan-to-client mapping. In practice, availability can therefore depend on the user’s plan, client, policy configuration, and rollout status.
The only Gemini 4 Argon-specific client requirement identified in the supplied documentation is for Visual Studio: the supported-model table lists Visual Studio 2022 version 17.14.22 or 18.1.0 as the minimum shown. Requirements for Visual Studio Code, JetBrains, Xcode, Eclipse, and Copilot CLI remain undocumented in the available material.
How to select the model
For supported IDEs, GitHub’s general model-selection guidance uses the Copilot Chat model picker. In Visual Studio Code, open Copilot Chat, select the current-model dropdown, and choose an available model. In Visual Studio, open GitHub Copilot Chat through the View menu and use the current-model dropdown.
In JetBrains IDEs, Eclipse, and Xcode, open Copilot Chat and select an alternative model from the model picker. The exact interface path differs by client, and the general documentation does not provide Gemini 4 Argon-specific instructions for each surface.
Changing the Copilot Chat model does not change the model used for inline suggestions. The supplied documentation also does not provide a Gemini 4 Argon-specific command or flag for Copilot CLI, so CLI users should not assume that the IDE selection procedure applies there.
GitHub’s general model-selection documentation covers the supported IDE workflows. If Gemini 4 Argon is not listed, the absence may reflect gradual rollout, plan availability, client support, or an administrator policy rather than a configuration error.
Administrator controls for Business and Enterprise
Business and Enterprise administrators can control access to AI models through organization or enterprise model-availability and policy settings. An administrator may therefore need to enable Gemini 4 Argon before members can select it.
For individual plans, GitHub’s configuration guidance says models can be used without organization or enterprise access configuration, although the models available to an individual depend on the plan. Administrators should review their existing model policies if the new option does not appear for users who otherwise meet the plan requirements.
GitHub’s AI model access documentation explains the organization and enterprise controls. The announcement’s surface list and the supported-model catalog should be read together: the announcement identifies the intended eligible plans and surfaces, while the extracted catalog tables do not independently confirm every individual plan-client combination.
Pricing and AI credit usage
Gemini 4 Argon uses Copilot’s token-based accounting for input, cached input, and output. GitHub converts usage into AI credits at a rate of one AI credit per $0.01 USD.
The listed promotional rates through December 1, 2026 are:
- $2.00 per 1 million input tokens
- $0.10 per 1 million cached input tokens
- $10.00 per 1 million output tokens
Individual plans include plan-dependent AI credit allowances. For Business and Enterprise, allowances are pooled at the billing-entity level. Usage beyond included allowances is billed at the applicable token rates.
These rates are promotional and should not be treated as the long-term price. Pricing after December 1, 2026 is not documented in the supplied sources. Teams evaluating the model should monitor token consumption, particularly for agentic tasks that may generate substantial output or repeatedly inspect project context.
Code completions and next-edit suggestions are not billed in AI credits and remain unlimited for paid Copilot plans under their existing counting mechanism. That treatment is separate from Copilot Chat and other model interactions that consume tokens.
What developers should use it for
Gemini 4 Argon is most clearly aimed at workloads where reasoning and multi-step coding assistance matter: complex implementation tasks, agentic work, and problems that require examining edge cases. It may be useful when a task needs more than a short completion or straightforward conversational answer.
GitHub’s early-testing statement suggests improvement over previous Gemini models, especially in Copilot CLI, but the absence of benchmark details makes it difficult to determine which tasks benefit most or how the model compares with other Copilot choices. Developers should evaluate it against their own repositories, tests, and review standards rather than assuming that the product description predicts every workflow.
What you should do
- Check the Copilot Chat model picker in your supported IDE during the gradual rollout.
- Confirm that your subscription is Copilot Pro+, Max, Business, or Enterprise.
- If you use Business or Enterprise, ask an organization or enterprise administrator to review model-availability policies.
- For Visual Studio, check that the installation meets the listed Gemini 4 Argon minimum of Visual Studio 2022 version 17.14.22 or 18.1.0.
- Review AI credit allowances and the promotional token rates before using the model extensively.
- Evaluate generated changes with your normal tests and human review; GitHub’s description of work checking is not a reliability guarantee.
Availability and remaining limits
Although GitHub describes Gemini 4 Argon as generally available, the rollout is gradual and simultaneous availability across every listed surface is not established. Complete compatibility requirements for several clients, Copilot CLI selection commands, context-window size, rate limits, and other model limits are also not documented in the supplied sources.
GitHub’s supported-model documentation says default AI model prompts and outputs run through Copilot content filters for harmful, offensive, or off-topic content, as well as public-code matching when enabled. That is a general Copilot behavior and should not be interpreted as a Gemini 4 Argon-specific security guarantee.
For the latest plan, surface, and billing details, consult GitHub’s announcement and the supported AI model catalog before standardizing the model across a team.
