GitHub moves Copilot to GPT-5.3

GitHub moved Copilot Business and Copilot Enterprise to GPT-5.3-Codex as their base model on May 17, according to a company changelog published that day. The change means enterprise customers that rely on the default Copilot setup are now routed to a newer coding model unless administrators have made other choices in policy or model settings. (github.blog) GitHub said the switch applies only to Copilot Business and Copilot Enterprise, not to Copilot Pro, Copilot Pro+, or Copilot Free. (openai.com) OpenAI, GitHub’s parent company, said separately that ChatGPT now has more than 900 million weekly active users. That figure, published in OpenAI posts this spring, places the GitHub change alongside a period of rapid consumer adoption for the company’s broader AI products. (github.blog) ### What exactly changed inside Copilot on May 17? May 17 was the “enablement” date for GPT-5.3-Codex as the base model for all Copilot Business and Copilot Enterprise organizations, GitHub said. In GitHub’s model system, the base model is the default model used when no other model is enabled or selected. (github.blog) March 18 was the date GitHub said it designated GPT-5.3-Codex as both the base model and the long-term support, or LTS, model for enterprise Copilot plans. GitHub’s documentation says customers then had a 60-day upgrade window before the new base model was automatically enabled. (openai.com) ### Which GitHub customers are covered, and who is not? Copilot Business and Copilot Enterprise are the only plans covered by the base-model change, GitHub said. The company said the change does not apply to Copilot Pro, Copilot Pro+, or Copilot Free, where model deprecation follows the standard timeline for individual plans. (github.blog) GitHub had already made GPT-5.3-Codex generally available earlier this year in Copilot chat surfaces including github.com, GitHub Mobile, Visual Studio Code and Visual Studio. For business and enterprise customers, administrators had to enable the GPT-5.3-Codex policy for users to see it in the model picker before the base-model switch took effect. (docs.github.com) ### Why does the “base model” matter more than a model-picker option? GitHub’s documentation says the base model is the default AI model Copilot uses when no other model is enabled. That matters because default settings shape what most employees encounter in day-to-day use, especially in managed enterprise environments where administrators control policies and extension versions. (github.blog) The same documentation says the base model carries a 1x premium request multiplier on paid plans. GitHub also said GPT-4.1 will remain force-enabled at a 0x multiplier “for the time being,” but will deprecate alongside the launch of usage-based billing on June 1, 2026. (github.blog) ### How does this fit with OpenAI’s broader scale? OpenAI said in March that ChatGPT had more than 900 million weekly active users and more than 50 million consumer subscribers. In a separate post last month, the company repeated the 900 million weekly-user figure and said enterprise accounts for more than 40% of its revenue. (docs.github.com) OpenAI also said Codex had reached 3 million weekly active users. The figures were published by OpenAI in corporate posts rather than securities filings, but they show the company presenting consumer usage and enterprise deployment together as it expands coding products across workplace software. (github.blog) ### What are the next dates enterprise customers need to watch? June 1, 2026 is the next operational deadline GitHub named in its changelog, when GPT-4.1 is set to deprecate alongside usage-based billing. February 4, 2027 is the end of the LTS availability window GitHub listed for GPT-5.3-Codex. (openai.com) GitHub told customers with questions or timing concerns to contact their account teams. The company’s Copilot documentation and changelog pages remain the main source for model status, billing changes and support dates for Business and Enterprise administrators. (github.blog) (openai.com)