{"meta":{"title":"Base and long-term support (LTS) models","intro":"Learn about base models, long-term support (LTS) models, and how they affect model availability for enterprises using GitHub Copilot.","product":"GitHub Copilot","breadcrumbs":[{"href":"/en/copilot","title":"GitHub Copilot"},{"href":"/en/copilot/concepts","title":"Concepts"},{"href":"/en/copilot/concepts/fallback-and-lts-models","title":"Base and LTS models"}],"documentType":"article"},"body":"# Base and long-term support (LTS) models\n\nLearn about base models, long-term support (LTS) models, and how they affect model availability for enterprises using GitHub Copilot.\n\n## About base models\n\n> \\[!IMPORTANT]\n>\n> * On March 18, 2026, GitHub designated GPT-5.3-Codex as the base model.\n> * Base models apply only to Copilot Business and Copilot Enterprise.\n\nA base model is the default AI model that GitHub Copilot uses when no other models are enabled. The base model is automatically enabled for all Copilot Business or Copilot Enterprise accounts within 60 days after the model is designated as a base model.\n\nWhen a new model is designated a base model, the following timeline applies:\n\n| Phase          | Timeline        | What happens                                                                                   |\n| -------------- | --------------- | ---------------------------------------------------------------------------------------------- |\n| Announcement   | Day 0           | GitHub announces the new base model.                                                           |\n| Upgrade window | Day 0 to Day 60 | Customers have 60 days to upgrade their IDE extensions to versions that support the new model. |\n| Enablement     | Day 60          | The new model is automatically enabled on all organizations and enterprises as the base model. |\n\n> \\[!NOTE]\n> The base model has a **1x premium request multiplier** on paid plans. For more information about multipliers, see [Requests in GitHub Copilot](/en/copilot/concepts/billing/copilot-requests#model-multipliers).\n\n## About long-term support (LTS) models\n\n> \\[!IMPORTANT]\n>\n> * On March 18, 2026, GitHub designated GPT-5.3-Codex as the LTS model.\n> * LTS models apply only to Copilot Business and Copilot Enterprise.\n\nAn LTS model is an AI model that GitHub commits to supporting for an extended period of one year from its designation date. During that period, the model remains available, which allows users to build around the model without concern that it will be closing down unexpectedly.\n\n## Continuous access when premium requests are unavailable\n\nGPT-5.3-Codex is available on paid plans with a 1x premium request multiplier, which means it consumes premium requests. To ensure continuous access to Copilot when premium requests are unavailable:\n\n* Premium request quota exhausted: If a user has used their monthly premium request allowance, Copilot automatically falls back to GPT-4.1 (the former base model) at no additional cost to the user.\n* Overage controls disabled: If an organization or enterprise has disabled premium request overages and a user reaches their limit, Copilot will fall back to GPT-4.1.\n\n## Further reading\n\n* [Supported AI models in GitHub Copilot](/en/copilot/reference/ai-models/supported-models)\n* [Requests in GitHub Copilot](/en/copilot/concepts/billing/copilot-requests)"}