Products & Announcements

OpenAI to Retire GPT-4o and Legacy ChatGPT Models on Feb 13, 2026

Corporate AI StrategyAI SafetyAI Ethics
Added Jan 29
Link415
Article: NeutralCommunity: NegativeDivisive
OpenAI to Retire GPT-4o and Legacy ChatGPT Models on Feb 13, 2026

OpenAI will remove GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini from ChatGPT on February 13, 2026; the API remains unchanged. GPT-4o’s strengths and user-favored style have been incorporated into GPT-5.1 and GPT-5.2 with new personality and customization controls. With most users on GPT-5.2, OpenAI is focusing on creativity and personality improvements, reducing unnecessary refusals, and building an adult-oriented experience with safeguards, including age prediction for minors.

Key Points

  • On February 13, 2026, GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini will be retired from ChatGPT; the API is unaffected for now.
  • GPT-4o was temporarily restored during the GPT-5 release based on user feedback valuing its warmth and creative ideation.
  • Feedback from GPT-4o users shaped GPT-5.1 and GPT-5.2, which now offer improved personality and customizable tone, warmth, and enthusiasm.
  • Retirement proceeds because capabilities are in place in newer models and usage has largely shifted to GPT-5.2, with only 0.1% still choosing GPT-4o daily.
  • OpenAI is reducing unnecessary refusals and preachy responses, aiming for an adult-oriented experience with safeguards, including age prediction for minors.

Sentiment

The community is predominantly critical of OpenAI's decision. While some accept model retirement as inevitable, the prevailing sentiment combines frustration over perceived quality regression in GPT-5 models, skepticism about OpenAI's user-preference framing, and concern for vulnerable users in the AI companion community. Many commenters express that they have already migrated to Claude or Gemini, suggesting the retirement is accelerating customer churn rather than consolidating satisfaction.

In Agreement

  • Retiring older models is a normal part of product evolution, and consolidating to fewer models allows OpenAI to focus improvement efforts
  • OpenAI has added personality customization controls (warmth, enthusiasm settings) that address concerns about losing 4o's conversational style
  • The 0.1% daily usage figure validates that the vast majority of users have successfully transitioned to newer models
  • Age verification and treating adults as adults is a reasonable approach to content moderation

Opposed

  • GPT-5/5.1/5.2 represent clear regressions from 4o and 4.1 in creativity, instruction following, conciseness, and hallucination rates
  • The 0.1% usage claim is misleading because 5.2 is the non-changeable default, requiring multiple clicks to switch to 4o each session
  • Retiring models that users depend on for production workflows and emotional support without adequate replacements is irresponsible
  • OpenAI should release retired model weights as open source rather than simply deleting them
  • The sycophancy problem is being reinforced by revealed-preference optimization rather than being addressed, exploiting users' impulse-control weaknesses
  • Users who built workflows, prompts, and even emotional relationships around specific model behaviors face real disruption with no recourse