Farewell, GPT-4o: OpenAI Retires Its Most Beloved Model on February 13
OpenAI is pulling the plug on GPT-4o in ChatGPT — and this time, it's for real. After a rocky first attempt during the GPT-5 launch and a user-driven reversal, the model that defined conversational AI for millions is finally heading into retirement.
What's Happening
On February 13, 2026, OpenAI will retire GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini from ChatGPT. Conversations will default to GPT-5.2 going forward. The API remains unaffected — for now.
Business, Enterprise, and Edu customers get a brief reprieve: GPT-4o stays available in Custom GPTs until April 3, 2026. After that, it's gone across all plans.
Why GPT-4o Mattered
GPT-4o wasn't just another model iteration. Launched in May 2024, it was OpenAI's first truly multimodal model — handling text, vision, and audio natively. But what made it special wasn't the spec sheet. It was the feel.
Users consistently described GPT-4o as warmer, more creative, and more conversationally natural than its successors. When OpenAI first tried to sunset it during the GPT-5 rollout, the backlash was immediate. Creative writers, educators, and daily users pushed back hard enough that OpenAI reversed course and restored access.
That feedback wasn't ignored. OpenAI says it directly shaped GPT-5.1 and GPT-5.2, with improvements to personality, creative ideation support, and new customization options for tone and style.
The Numbers Tell the Story
Despite the emotional attachment, usage data paints a clear picture: only 0.1% of daily users still choose GPT-4o. The vast majority have already migrated to GPT-5.2, which now offers the personality controls that made GPT-4o beloved in the first place.
What's NOT Changing
Two important clarifications from OpenAI:
- Voice Mode continues unchanged. While it uses a similar base model, it's technically separate from the text GPT-4o being retired.
- Image Generation also stays. Same logic — different model under the hood.
What This Means for Developers
If you're building on the OpenAI API, breathe easy. There are no API changes tied to this announcement. The chatgpt-4o-latest endpoint has a separate deprecation timeline (February 17, 2026), but the core GPT-4o API models remain available.
That said, the writing is on the wall. OpenAI is consolidating around GPT-5.2 as its flagship, and API deprecations for older models will follow eventually.
The Bigger Picture
This retirement is part of OpenAI's broader model consolidation strategy. In the space of a few months, they're clearing out GPT-4o, GPT-4.1, GPT-4.1 mini, o4-mini, and the GPT-5 Instant and Thinking variants — all in favor of GPT-5.2.
It's a bold move that signals confidence in their latest model, but also reflects a lesson learned: you can't just yank models away without giving users time to adapt and alternatives that actually address their needs.
GPT-4o earned its place. It changed how people think about talking to AI. And while GPT-5.2 may be objectively more capable, there's something to be said for the model that made millions of people feel like they were having a real conversation for the first time.
Farewell, GPT-4o. You were one of the good ones.