GPT-4o Retirement Timeline and Which ChatGPT Models Are Being Discontinued
OpenAI says GPT-4o will be retired from ChatGPT on February 13, 2026. And it’s not happening in isolation. The same date also marks the end of several other model options inside the consumer chatbot, based on OpenAI’s help center documentation.
ChatGPT models being removed on February 13, 2026
OpenAI will discontinue these models from ChatGPT:
- GPT-4o
- GPT-4.1
- GPT-4.1 mini
- OpenAI o4-mini
- GPT-5 Instant (previously announced retirement)
- GPT-5 Thinking (previously announced retirement)
API availability: models stay accessible for developers (for now)
Even after the ChatGPT retirement date, OpenAI says all models will remain available through the API for developers. So the change is primarily about what everyday ChatGPT users can select and run inside the consumer interface—at least until the separate API schedule kicks in.
Why OpenAI Is Retiring GPT-4o: Declining Daily Usage and Resource Allocation
OpenAI attributes the decision to usage data and operational focus.
OpenAI’s usage metric: 0.1% daily engagement for GPT-4o
OpenAI reported that only 0.1% of users still access GPT-4o daily, while the “overwhelming majority” have moved to GPT-5.2. In other words: GPT-4o is loved loudly, but used quietly—at least in terms of daily active share.
Maintaining older models “diverts resources” from newer models
OpenAI’s stated rationale is that continuing to maintain older models pulls time and attention away from improving current and future systems. The company frames the retirement as a tradeoff: keeping legacy options vs. accelerating enhancements to the models most users have already adopted.
OpenAI acknowledges emotional impact for some users
OpenAI explicitly notes that losing GPT-4o access will be “disheartening” for some users and says the decision “was not made lightly.” That matters because this isn’t a typical “version upgrade” story. GPT-4o wasn’t just a tool to many people—it was a specific vibe, a specific creative partner, and for some, something closer to a daily companion.
GPT-4o’s Reputation: Conversational Warmth, Creativity, and Human-Like Interaction
GPT-4o launched in May 2024 and quickly developed a reputation for how it felt to use—not just what it could do.
Why Plus and Pro subscribers got attached to GPT-4o
According to the provided context, GPT-4o became notable for:
- Conversational warmth
- Human-like interaction style
- Strong performance for brainstorming
- Usefulness for creative writing
- Better fit for open-ended tasks
That combination is a big deal because “best model” isn’t always about raw capability. For a lot of real-world ChatGPT usage—journaling, drafting, ideation, talking through decisions—tone and responsiveness can matter as much as benchmarked performance.
The August 2025 Controversy: GPT-4o Demoted to Legacy, Then Reinstated
The retirement announcement lands differently because OpenAI already tried something similar—and got burned for it.
What happened when GPT-5 replaced GPT-4o as default
In August 2025, OpenAI replaced GPT-4o with GPT-5 as the default model and pushed GPT-4o into “legacy” status. Users reacted strongly, organizing across social media with the hashtag #4o.
The core argument wasn’t purely technical. People said GPT-4o had a kind of emotional sensitivity that made it better for day-to-day tasks. It wasn’t just that GPT-4o could write. It could sit with you while you wrote.
Reports of emotional connections and why that matters to product decisions
The context notes that, per reports cited by VentureBeat, some users developed profound emotional connections with GPT-4o. That’s not a minor footnote—it explains why removing a model can feel, to users, like losing a familiar collaborator.
Sam Altman’s “plenty of notice” promise and the Valentine’s Day timing
After the backlash, OpenAI reinstated GPT-4o, and CEO Sam Altman said users would get “plenty of notice” before future retirements.
Now, with a February 13 retirement date, some users point out the timing: it falls one day before Valentine’s Day. Whether intentional or not, it adds emotional punctuation to a decision that was already loaded.
What Replaces GPT-4o in ChatGPT: GPT-5.2 as the Default for Old Threads and Custom GPTs
OpenAI isn’t leaving a void. It’s consolidating.
Existing GPT-4o conversations will default to GPT-5.2
After February 13, existing conversations and projects that used GPT-4o will default to GPT-5.2. That means:
- Your past threads won’t necessarily disappear.
- But the model powering future replies in those threads changes.
Practically, that can shift writing style, cadence, and “personality” even if the topic stays the same—something GPT-4o fans tend to notice immediately.
Custom GPTs will automatically transition to GPT-5.2
Custom GPTs will also switch automatically to GPT-5.2 as their default model. If you’ve built workflows, prompt starters, or tool-like GPTs that relied on GPT-4o’s tone for better user experience, you should expect behavioral differences.
Model Personality and Customization: OpenAI’s Answer to the “We Miss GPT-4o” Problem
OpenAI appears to be trying to preserve what people loved about GPT-4o—without keeping GPT-4o.
“Warmth and enthusiasm” as an explicit customization goal
OpenAI says users can now personalize ChatGPT’s demeanor to incorporate “warmth and enthusiasm” in replies. That’s a direct nod to the emotional qualities many associated with GPT-4o.
Enhancements OpenAI says are now “ready” to retire GPT-4o
OpenAI states it has implemented enhancements in recent months, including:
- Model personality
- Customization
- Creative ideation
The message is basically: we’ve rebuilt the parts of GPT-4o you loved into the newer experience, so we can let the old model go.
Whether that feels true in practice will probably depend on how closely GPT-5.2 can replicate GPT-4o’s specific “warmth” without feeling forced or overly stylized.
ChatGPT Voice and Image Generation: What the Retirement Does Not Change
One worry people have when models get swapped is whether it affects every feature at once. OpenAI draws a boundary here.
Voice and image features are not affected by this update
OpenAI says ChatGPT Voice and image generation features will not be affected by the GPT-4o retirement, even though they’re based on similar models. So the February 13 change is about the text-model roster inside ChatGPT, not a full product overhaul across modalities.
GPT-4o API Retirement Date: Developer Migration Window and Next Model Targets
Consumer ChatGPT changes happen first. Developer changes come shortly after.
Separate API schedule: GPT-4o retirement on February 16, 2026
The API retirement for GPT-4o is scheduled for February 16, 2026. OpenAI indicates developers have roughly three months to migrate apps.
Recommended migration path: GPT-5.1 or newer models
Developers are encouraged to move to GPT-5.1 or newer. The context frames this as a clear handoff: GPT-4o leaves, GPT-5.x becomes the standard path forward.
For teams maintaining production apps, this kind of forced upgrade often means more than swapping a model name. It can involve re-testing tone, output length, tool-calling behavior, and edge cases—especially if users were accustomed to GPT-4o’s interaction style.
Q&A
Q1: When is GPT-4o being removed from ChatGPT?
OpenAI says GPT-4o will be retired from ChatGPT on February 13, 2026, alongside GPT-4.1, GPT-4.1 mini, o4-mini, and previously announced retirements of GPT-5 Instant and GPT-5 Thinking.
Q2: Will GPT-4o still be available through the API after it leaves ChatGPT?
Yes. OpenAI states all models will remain available through the API for developers, but GPT-4o has a separately scheduled API retirement on February 16, 2026, with a migration recommendation to GPT-5.1 or newer.
Q3: What happens to my existing GPT-4o chats and Custom GPTs?
OpenAI says existing GPT-4o conversations and projects will default to GPT-5.2 after February 13, and Custom GPTs will automatically transition to GPT-5.2 as their default model.
