OpenAI Retired GPT-4o From ChatGPT in February
OpenAI announced on January 29 that GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini would leave ChatGPT's main model picker on February 13, 2026. OpenAI said only 0.1% of users still selected GPT-4o daily and that API access continued; its current help page says Business, Enterprise, and Edu workspaces retained GPT-4o in Custom GPTs until April 3.
OpenAI removed GPT-4o and several other legacy models from ChatGPT's main model picker on February 13, 2026, following a January 29 announcement. The change also covered GPT-4.1, GPT-4.1 mini, and o4-mini, alongside previously announced GPT-5 retirements.
This is an archive story about a completed February product change, not a July launch. OpenAI's current help documentation confirms the main ChatGPT retirement, says the affected models continued through the API, and records a narrower exception: Business, Enterprise, and Edu workspaces retained GPT-4o within Custom GPTs until April 3.
Why OpenAI made the change
OpenAI said usage had shifted heavily toward GPT-5.2 and that only 0.1% of users were still choosing GPT-4o each day. The company also said feedback from people who preferred GPT-4o's conversational style had influenced personality, creativity, and customization work in newer models.
The decision followed an earlier attempt to remove GPT-4o after the GPT-5 launch. OpenAI restored access after user backlash, then gave advance notice before the February retirement. Independent reporting documented strong resistance from users who had built creative workflows or emotionally significant relationships around GPT-4o.
What changed for users
The February 13 change removed the listed models from ChatGPT's main model picker. It did not retire the base models from the API at that time. OpenAI's help page also treats ChatGPT Voice and ChatGPT Images as separate systems that were not changed by the text-model retirement.
The product-specific exception matters: GPT-4o remained available inside Custom GPTs for Business, Enterprise, and Edu workspaces until April 3. That does not change the February model-picker date, but it means access did not end identically across every ChatGPT surface on February 13.
For teams that depended on a particular model's behavior, the episode is a reminder that a hosted product interface is not a permanent runtime contract. Important workflows should document model assumptions, maintain regression tests, and plan for migrations when providers change the available model set. That is LDS interpretation; OpenAI's stated reason was to focus development on the models most people were using.
Key Points
- 1OpenAI announced the retirement on January 29 and removed GPT-4o and other legacy models from ChatGPT's main model picker on February 13, 2026.
- 2OpenAI said only 0.1% of users still selected GPT-4o daily as usage shifted to GPT-5.2.
- 3API access continued, while Business, Enterprise, and Edu workspaces retained GPT-4o inside Custom GPTs until April 3.
Scoring Rationale
A completed product retirement with meaningful workflow and user-continuity implications, but the exact event occurred in February and is presented as archive context.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
