On 2 October 2026 GitHub announced that a set of Copilot models – including Gemini 3.5 Flash, Gemini 3.6 Flash, Gemini 3.8 Flash, Kimi K2.7 Code, Claude Opus 4.7 and Claude Opus 5.5 – are deprecated across every Copilot surface (Chat, inline edits, ask/agent modes, and code completions). Any pipeline, IDE extension, or automation that references these identifiers will start receiving errors unless it is switched to a supported model.
Deprecation specifics
Gemini 3.5 Flash– deprecated 2026‑10‑02, suggestedGemini 3.8 FlashGemini 3.6 Flash– deprecated 2026‑10‑02, suggestedGemini 3.8 FlashKimi K2.7 Code– deprecated 2026‑10‑02, suggestedKimi K3Claude Opus 4.7– deprecated 2026‑10‑02, suggestedClaude Opus 5.5
Operational impact
All existing Copilot integrations that hard‑code the deprecated model names must be updated to reference the suggested alternatives. The change does not require manual removal of the old models; GitHub will simply stop serving them. However, Copilot Enterprise administrators may need to adjust model‑policy settings to grant access to the new models. After enabling a model, it appears in the Copilot Chat selector in VS Code and on github.com, and can be verified in the individual Copilot settings page.
Architectural considerations
Teams that embed Copilot calls in CI/CD jobs, custom bots, or internal tooling should audit their configuration files (e.g., .copilotrc or API request payloads) for the deprecated identifiers. Switching to the newer models may affect output style or token usage, so a brief validation run is advisable before promoting the change to production. Because model access is governed by policy, ensure that the policy update propagates to all relevant service accounts or runner environments.
Related CloudNinjas coverage: AI engineering.
What This Means For Practitioners
- Inventory every place where a Copilot model name is specified.
- Replace deprecated identifiers with the suggested alternatives.
- Confirm that the alternative model is enabled in your organization’s Copilot model policy.
- Run a smoke test in a non‑production environment to verify that the new model meets functional expectations.
- If you encounter policy‑related blocks, work with your account manager to adjust the enterprise settings.


