Grok 4.5 is gradually rolling out in GitHub Copilot

GitHub Changelog announced on July 28, 2026 that xAI Grok 4.5 is gradually rolling out in GitHub Copilot. The primary source describes Grok 4.5 as designed for fast agentic coding and complex multi-step workflows, with a context window of up to 500,000 tokens, text and image input support, and low, medium, and high reasoning effort settings.

GitHub says its internal testing showed strong results on terminal-based coding tasks in Visual Studio Code and Copilot CLI, especially when dispatching tools in parallel and taking direct action. That is GitHub primary-source information, not an independently verified benchmark. It does show how Copilot model selection is extending beyond completion into CLI and agent-oriented workflows.

GitHub Docs also lists Grok 4.5 as an xAI-provided GA model in Supported AI models in GitHub Copilot. That supporting documentation aligns with the changelog announcement.

Grok 4.5 enters the model picker for Pro through Enterprise plans

According to the primary source, Grok 4.5 will be available to Copilot Pro, Pro+, Max, Business, and Enterprise SKUs. GitHub lists Visual Studio Code, Visual Studio, Copilot CLI, GitHub Copilot cloud agent, the GitHub Copilot app, JetBrains, Xcode, and Eclipse as places where users will be able to select it in the model picker.

The rollout is gradual. Even on an eligible plan, the model may not appear for every user immediately on July 28, 2026. GitHub Docs explains that model availability depends on the Copilot plan, the client being used, and whether an organization or enterprise restricts access to particular models.

If the model does not appear, separate the checks by plan, client, organizational policy, and rollout timing. Those causes look similar from the model picker, but they lead to different actions.

Business and Enterprise admins must enable the Grok 4.5 policy

The organization-specific detail is important: the primary source says Copilot Enterprise and Copilot Business administrators must enable the Grok 4.5 policy in Copilot settings, and that the policy is off by default.

GitHub Docs on Configuring access to AI models in GitHub Copilot says access to Copilot models depends on the plan, client, and whether an organization or enterprise restricts specific models. It also says individual plans can generally use models directly, while Free and Student users only have access through auto model selection.

For administrators, the update is therefore not just a model-picker addition. It is a policy decision covering who can use the model, where it should be tested, and how it fits existing AI usage controls.

The 500,000-token context and reasoning effort settings affect cost

The primary source says Grok 4.5 is billed at provider list pricing under usage-based billing. GitHub Docs on Models and pricing for GitHub Copilot explains that Copilot interactions consume input, output, and cached tokens, and that cost depends on both the selected model and the number of tokens consumed. The same page defines 1 AI credit as 0.01 USD.

On the provider side, Models | SpaceXAI Docs lists grok-4.5 with a 500k-token context window, input pricing of 2.00 USD per 1M tokens, output pricing of 6.00 USD per 1M tokens, and configurable reasoning. GitHub Copilot billing still follows the GitHub billing system, but this provider page helps explain the pricing reference in the changelog.

Long context and higher reasoning effort can be useful for complex tasks, but they can also increase token consumption. GitHub supported-models documentation says larger context windows and higher reasoning levels affect AI credit consumption and recommends regular context and reasoning by default, reserving larger settings for more complex work.

xAI announced Grok 4.5 before the Copilot integration

For background, xAI published Introducing Grok 4.5 on July 16, 2026, positioning the model for coding, agentic tasks, and knowledge work. That announcement also says Grok 4.5 was available in Grok Build, Cursor, and the SpaceXAI console, and it includes an API example using grok-4.5.

The July 28, 2026 GitHub Changelog is therefore not the initial launch of the model itself. It is the event that brings Grok 4.5 into GitHub Copilot plan, client, model-picker, and organization-policy framework.

As related history, GitHub Docs records Grok Code Fast 1 as retired on May 15, 2026 in the Copilot model retirement table. The primary source does not describe Grok 4.5 as its direct replacement, so that should remain background context rather than a claimed migration path.

Checks before trying Grok 4.5 in Copilot

Developers should first check whether their Copilot plan and client expose Grok 4.5 in the model picker. In Business or Enterprise environments, a missing model may come from rollout timing, an administrator policy, or client support rather than from local setup alone.

Administrators should decide whether to enable the default-off Grok 4.5 policy after reviewing target users, supported clients, AI credit handling, and the likely cost profile of long-context or higher-reasoning work. As the model list grows, teams also benefit from lightweight guidance on which model to use for which type of task.

Finally, GitHub internal testing statements should be treated as GitHub evaluation. Before making Grok 4.5 a default choice, test it on real team workflows: repository exploration, CLI-driven tasks, multi-file edits, and image-input cases where relevant.

Source

• Title: Grok 4.5 is now available in GitHub Copilot
• URL: https://github.blog/changelog/2026-07-28-grok-4-5-is-now-available-in-github-copilot