If you blinked, you missed three ChatGPT updates. That's not an exaggeration — it's roughly the actual cadence.
The biggest change in March: GPT-5.1 models are gone. As of March 11th, GPT-5.1 Instant, GPT-5.1 Thinking, and GPT-5.1 Pro have been retired from ChatGPT entirely. Existing conversations that used those models automatically continue on their successors — GPT-5.3 Instant, GPT-5.4 Thinking, and GPT-5.4 Pro respectively. If you hadn't noticed the switch, that's probably a sign that the transition was smooth. Or that the differences between versions are getting small enough that most users can't distinguish them.
GPT-5.4 mini is rolling out as a thinking model available to Free and Go tier users, and as a rate-limit fallback for paid users. It's lighter, faster, and designed to handle the kind of reasoning tasks that previously required the full-weight models. The implication is clear: OpenAI wants thinking capabilities available at every price point, including free.
“GPT-5.4 mini is rolling out as a thinking model available to Free and Go tier users, and as a rate-limit fallback for paid users.”
In the personality department, the Nerdy preset — one of ChatGPT's base style options — is being discontinued. Users who selected it will be moved to the default personality. The stated reason is simplification, but reading between the lines, it feels like OpenAI realized that multiple personality presets created quality inconsistencies that made the product harder to evaluate and improve.
Key Takeaways
- →ChatGPT: GPT-5.
- →OpenAI: GPT-5.
- →AI: GPT-5.
- →GPT-5: GPT-5.
A tone improvement to GPT-5.3 Instant specifically targeted the follow-up responses that sounded like clickbait — the "You'll never believe…" and "If you want to know more…" phrasings that made the model sound like a pushy content creator rather than an assistant. Small fix, big quality-of-life improvement.
Legacy deep research mode gets deprecated on March 26th. The current deep research feature stays.
On the business side, the more consequential story: OpenAI is actively preparing for an IPO, potentially before the end of 2026. CEO of Applications leadership is emphasizing enterprise productivity use cases — signaling that the company views its future revenue in business tools, not consumer chatbot subscriptions.
The pace of change at OpenAI is something you either admire or find exhausting. Probably both.