On October 7, 2026, GitHub announced that Claude Haiku 5.5 is now generally available in GitHub Copilot, after a gradual rollout period. The model joins the model picker across every Copilot surface, for Pro, Pro+, Max, Business, and Enterprise subscriptions.
A broad rollout, surface by surface
Claude Haiku 5.5 is available from Visual Studio Code, Visual Studio, Copilot CLI, the GitHub Copilot cloud agent, the Copilot app, github.com, GitHub Mobile (iOS and Android), JetBrains IDEs, Xcode, and Eclipse. The rollout stays gradual according to GitHub: some accounts may not yet see the model appear in their picker.
For Business and Enterprise organizations, access is managed through the model policy in Copilot settings. By default, a new model is enabled automatically unless an administrator has disabled global default enablement or this specific model.
| Criterion | Detail |
|---|---|
| Announcement date | October 7, 2026 |
| Status | General availability |
| Plans covered | Pro, Pro+, Max, Business, Enterprise |
| Billing | Provider list pricing, under usage-based billing |
| Admin control | Copilot model policy (enable/disable per model) |
Billing: the checkbox that matters
GitHub notes that the model is billed at the provider’s standard list pricing under usage-based billing, with no mention of a specific premium request multiplier on the announcement page. For an organization tracking its Copilot costs at the team level, the check is straightforward: verify in the model policy whether default enablement was left open, then watch consumption over the first weeks of real-world use.
# Vérifier les modèles actifs pour une organisation (exemple d'appel API)
gh api /orgs/ORG/copilot/billing/seats --jq '.seats[] | select(.assigning_team != null) | {login: .assignee.login, plan: .plan_type}'
Where Haiku 5.5 fits in the model lineup
A generally available Haiku model rounds out the lineup already present in Copilot, where several model families coexist for different use cases: fast inline completion, long-running reasoning on a cloud agent, or automated code review. Model choice remains manual through the picker, or governed by organization policy.
The model is now available across every Copilot surface, from VS Code to GitHub Mobile, with a rollout that varies by account.
An account that doesn’t yet see Claude Haiku 5.5 in its model picker likely has nothing to fix: GitHub is rolling out availability gradually, and the model’s absence today isn’t a sign of misconfiguration.
What this means for teams managing their AI spend
The addition fits a broader trend: as more models become available inside the same tool, teams increasingly need a formal policy for choosing between them, rather than leaving each developer to pick a model at random. Teams already running agents in production benefit from documenting which model is recommended for which type of task.
Key takeaways
Claude Haiku 5.5 leaves its gradual rollout phase to become generally available in Copilot, across nearly every surface the tool offers. Billing follows the provider’s standard list pricing, and activation control stays in the hands of organization administrators through the model policy.
Announcements like this one often go unnoticed even though they have a direct impact on the end-of-month bill: a model enabled by default, multiplied across an organization’s seats, can push consumption up without anyone making an explicit decision. Checking the model policy after every announcement like this should be routine, not a reaction after the invoice lands. — Simon Janvier
Further reading: the official announcement on the GitHub changelog.
