Gemini 4 Argon reaches GitHub Copilot with usage-based pricing

GitHub [announced on October 13](https://github.blog/changelog/2026‑10‑13‑gemini‑4‑argon‑in‑github‑copilot/) that Gemini 4 Argon is generally available in Copilot for Pro+, Max, Business and Enterprise users. The release covers Visual Studio Code, Visual Studio, Copilot CLI, the Copilot coding agent, the Copilot app, JetBrains IDEs, Xcode and Eclipse. The rollout is gradual. A qualifying subscription therefore does not prove the model has reached every account or client yet. GitHub tells users who do not see it to check again later. The practical test is the model picker in the client a team actually uses. GitHub's [supported‑model table](https://docs.github.com/en/copilot/reference/ai‑models/supported‑models) also warns that availability depends on the plan and surface, so an appearance in one client should not be treated as fleet‑wide access.

GitHub's [models and pricing reference](https://docs.github.com/en/copilot/reference/copilot‑billing/models‑and‑pricing) lists Gemini 4 Argon at $2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens through December 1, 2026. GitHub converts token cost into AI credits at one cent per credit. At those list rates, a request with 100,000 uncached input tokens and 10,000 output tokens is about $0.30, or 30 AI credits: $0.20 for input and $0.10 for output. That calculation excludes any additional work, retries or agent infrastructure and is an illustration rather than a prediction of a team's bill. Copilot plans include different AI‑credit allowances. Additional usage is billed after the applicable allowance, so administrators should compare token volume and output length rather than assuming that a GA model is included without limit.

Copilot Business and Enterprise administrators can manage Argon through the model policy in Copilot settings, according to the [release note](https://github.blog/changelog/2026‑10‑13‑gemini‑4‑argon‑in‑github‑copilot/). Under GitHub's default‑model enablement behavior, a new model is enabled automatically unless an administrator has disabled the global default or explicitly disabled that model. That makes this release a governance event as well as a model addition. An organization that wants explicit review of every new model should verify its global default before rollout reaches users. A team that wants Argon should check both the global rule and the model‑specific setting. Policy state also needs a client‑side check. Record the plan, client version, policy decision and whether Argon actually appears for a representative managed user.

GitHub says early testing found that Gemini 4 Argon solved more tasks than previous Gemini models on every coding benchmark it used, with the largest gains in Copilot CLI. The [one‑minute release note](https://github.blog/changelog/2026‑10‑13‑gemini‑4‑argon‑in‑github‑copilot/) does not publish benchmark names, scores, sample sizes or error bars. That statement supports a reason to evaluate the model, not a universal quality conclusion. Repository mix, tool permissions, context size, latency and output cost can change the result for a real team. A useful trial keeps the task set fixed, reviews accepted patches rather than raw completion counts, and records cost and time to an accepted result. Compare against the team's present model on the same repositories and guardrails.

Start with a small group and two or three repeatable coding jobs: a bug fix with tests, a repository explanation and a constrained refactor. Confirm the exact model label, retain prompts and diffs under the team's normal policy, and record human review or rework. Measure access failures, total token cost, latency and accepted outcomes separately. A faster response that requires more correction is not automatically cheaper; a stronger result that consumes more output tokens may still be worthwhile. The release expands choice inside Copilot, but it does not remove the need for model policy, client compatibility or task‑specific evaluation. Follow [GitHub's OpenTools organization page](https://opentools.ai/organizations/github) for material Copilot changes. *Image: GitHub's official Copilot model‑picker image showing Gemini 4 Argon. The source image was downloaded from the October 13 GitHub Changelog page without generative alteration or upscaling.*