🇸🇬Singapore·AI News·16 Nov 2026·via Maketecheasier

    GitHub is retiring six AI models from Copilot on 14 December, including Gemini 3.8 Flash and Grok 4.6, across chat, inline edits and code completions.

    GitHub is retiring six AI models from its Copilot service on 14 December 2026, including Gemini 3.8 Flash and Grok 4.6. The change, announced in a GitHub Changelog post on 13 November 2026, affects all Copilot experiences, from chat and inline edits to ask and agent modes and code completions.

    GitHub's decision to deprecate six AI models in Copilot by 14 December 2026 requires users to update their workflows and integrations. The models being retired are Claude Fable 5, Claude Haiku 4.5, Gemini 3.8 Flash, Gemini 4 Argon, Grok 4.6, and Kimi K3. This change applies across all Copilot functionalities, including chat, inline edits, agent modes, and code completions.

    For most retiring models, GitHub has provided suggested alternatives. For example, Fable 5 is succeeded by Fable 5.2, Gemini 3.8 Flash by Gemini 4 Hydrogen, Gemini 4 Argon by Gemini 4 Carbon, and Grok 4.6 by Grok 4.7. However, Kimi K3 currently has no announced successor, which could create a workflow gap for teams that have standardized on that model.

    A specific carve-out exists for Claude Haiku 4.5. While it is being deprecated for most customers, individual GitHub Copilot subscribers on annual plans will retain access. Other customers are directed to Claude Haiku 5.5. This distinction means that different user segments will experience the transition to new models in varied ways.

    For Copilot Enterprise and Copilot Business customers, the suggested alternative models are automatically enabled under certain conditions. This automatic enablement occurs unless an administrator has previously turned off the global default setting or explicitly disabled the specific model. In such cases, the replacement models will not appear automatically.

    Administrators managing Copilot for their organizations must understand how to enable new models where automatic enablement does not apply. GitHub states that if the global default is off, administrators can enable access to the alternative models through their model policies in Copilot settings, which is applied per model rather than via a single global switch.

    Users can verify which models are available to them through the Copilot Chat model selector within supported GitHub Copilot experiences. The announcement from GitHub did not include any information regarding changes to pricing, seat counts, or the overall plan structure, focusing solely on the model list update.

    Share this article

    Original reporting by MaketecheasierWe don't republish, read the full story â†’

    Related reading

    6 stories