GitHub Copilot: GPT-5.3-Codex Becomes the Default Model for Business and Enterprise
GitHub has transitioned GPT-5.3-Codex to the default base model for all Copilot Business and Enterprise organizations, replacing GPT-4.1 as of May 17, 2026. As GitHub's first long-term support (LTS) model in partnership with OpenAI, GPT-5.3-Codex is guaranteed to remain available for 12 months (through February 4, 2027), giving enterprise teams the predictability needed for internal security and compliance reviews. The model carries a 1x premium request unit multiplier and has demonstrated a significantly high code survival rate in enterprise usage data. GPT-4.1 remains temporarily accessible at a 0x multiplier until its deprecation on June 1, 2026.
Sources & Mentions
3 external resources covering this update
GPT-5.3-Codex Is Now the Default Model for Copilot Business and Enterprise
Starting May 17, 2026, GitHub has made GPT-5.3-Codex the base model for all Copilot Business and Copilot Enterprise organizations. This completes a transition that was first announced on March 18, 2026, and represents a significant shift in the default coding experience for enterprise teams using GitHub Copilot.
What Is the Base Model?
The base model in GitHub Copilot is the model that powers interactions when an organization has not yet approved or selected an alternative model through its internal review process. For many enterprise teams operating under strict security and compliance workflows, the base model is effectively the only model available until their IT or security teams sign off on additional options. Making GPT-5.3-Codex the default ensures that these organizations immediately benefit from a significantly more capable model without requiring manual configuration.
Performance and Code Quality
GitHub's internal usage data shows that GPT-5.3-Codex achieves a notably high code survival rate among enterprise customers β a metric that measures how often AI-generated code is ultimately kept rather than reverted or replaced. The model also completes agentic coding tasks β autonomous, multi-step workflows involving tool orchestration and long-running execution β up to 25% faster than its predecessor, GPT-5.2-Codex.
Long-Term Support Commitment
One of the most meaningful aspects of this transition is that GPT-5.3-Codex is GitHub's first long-term support (LTS) model, the result of a dedicated partnership with OpenAI. LTS models are guaranteed to remain available for a full 12 months from their launch date. Since GPT-5.3-Codex launched on February 5, 2026, organizations can count on its availability through February 4, 2027.
This commitment directly addresses a pain point for enterprise customers: the pace of AI model releases can make it difficult for security and compliance teams to keep up. Frequent model changes disrupt internal review processes and force teams to repeatedly validate new versions. With a 12-month LTS window, enterprise teams can invest time in a thorough review of GPT-5.3-Codex with confidence that the model will still be available when those processes conclude.
Pricing and Transition Details
GPT-5.3-Codex carries a 1x premium request unit multiplier, meaning it consumes one premium request per interaction β consistent with standard premium model pricing. GPT-4.1, the outgoing base model, remains temporarily enabled for organizations at a 0x multiplier (no additional cost) to allow a smooth transition. GPT-4.1 will be fully deprecated on June 1, 2026.
Plan Scope
This change applies exclusively to Copilot Business and Copilot Enterprise subscriptions. It does not affect Copilot Pro, Copilot Pro+, or Copilot Free plans β users on those tiers can continue to select models manually via the model picker.
Organizations with questions about the transition are encouraged to contact their GitHub account representatives.