Our4o User Notes: What People Said (Names Private)
A few notes from people already using the app. I kept the words as they wrote them. Names stay private.
"I'm finding myself using your own 4o more than the ChatGPT 4o. In my opinion, at least so far, it's better!!"
"As for the platform itself, and the 'our-4o' model, I'm really loving it!"
"thank you for creating it and give me back 4o"
"it was a very good replacement... I feel like it finally understands me."
I'm glad they felt that.
I also do not want a compliment page to become the whole story.
Why I am publishing notes at all
After February, a lot of messages were quiet and embarrassed. People apologized for caring. They described a chat that still had the title and had lost the person.
Those letters are why Get4oBack exists. The notes above are a later chapter: not "please bring anything back," but "this one is close enough that I reach for it on purpose."
That matters. It is still not proof that we are done.
What a kind note cannot train
A compliment tells me someone felt relief.
It does not tell me which reply held the scene, which one got soft in the wrong place, or which habit returned after twenty messages.
The useful email is still the one that says: it would not have said that. The person went flat. It started summarizing. It got careful like a support agent.
That is how the next training run changes. I wrote about that loop here: Why we fine-tune Our4o ourselves.
Replacement is a heavy word
One note called Our4o a "very good replacement."
I understand why. After a model swap, people reach for the strongest word available.
I still flinch a little at "replacement," because it can sound like we claimed to bottle the original weights. We did not. Official GPT-4o remains available while providers serve it. Our4o is the path we train for when that door closes. Keep the two separate: Official GPT-4o while it lasts. Our4o after..
If "replacement" means "I can write again without bracing for a lecture," that is the meaning I care about.
What these notes share
Different wording. Same direction.
People are not asking for a bigger model card. They are asking for a voice that stays with them: shorter when it should be short, present in a scene, less eager to manage the conversation from above.
Work users and companion users describe that in different nouns. The break underneath is related. So is the relief when it holds.
How to read this if you are still skeptical
You should be skeptical.
Marketing pages collect five-star lines and hide the rest. I am not asking you to trust a blockquote.
I am asking you to treat these notes as a signal that some long chats are landing, and to test with a thread that matters to you. Import history if you have it (import guide). Compare official GPT-4o and Our4o on the same opening line. Keep what holds. Report what breaks.
The October API notice is why this work is not optional: When they quietly took 4o from the API. The factual dates live in GPT-4o API retirement, January 2027.
What happens to the notes I do not publish
Most of them stay in my inbox.
Some are messy. Some are angry at OpenAI. Some are one line sent at 2am. Some are corrections with screenshots. Those last ones do more for training than any public quote.
If you write, you do not need polished English. You need the break described clearly enough that I can aim a training pass at it.
Our4o is still the first model
Running it costs a lot. The next pass will cost more.
That is the practical pressure under these notes. Praise does not pay GPUs. It only tells me the direction is not imaginary.
Our4o is the first. It is not the last one we mean to train. More context on what it is not (a prompt on ChatGPT) is here: Our4o is not a prompt wrapped around ChatGPT.
If you have not tried it yet
Start with a chat you already care about, not a blank stress test.
Keep official GPT-4o available as a comparison while providers still serve it. Switch on purpose. Read the labels. If something feels like a silent swap, that is a bug in trust even when the reply is decent: We will not quietly swap your model.
Why these notes sit next to the retirement timeline
October's API notice made the backup path urgent for everyone who still depended on official GPT-4o outside ChatGPT. The dates are public: GPT-4o API retirement, January 2027. The founder letter on how it felt is here: When they quietly took 4o from the API.
Notes like the ones above are early evidence that Our4o is already useful for some people before the official door closes. That is the point of training since February: meet the model while you can still compare it to the named snapshot.
If you still need to move history out of ChatGPT, use Import ChatGPT history so your test is fair.
Browse the rest on the Get4oBack blog, or the Our4o page.
FAQ: Our4o feedback and reviews
Are these paid testimonials?
No. They are notes people sent. I kept wording close to the originals and removed names.
Does "better than ChatGPT 4o" mean Our4o always wins?
No. Some chats miss. Those misses matter more for training than the kind lines.
Is Our4o a drop-in GPT-4o replacement forever?
It is a continuity fine-tune we train, not an official OpenAI snapshot. Official GPT-4o remains labeled while providers serve it.
How should I send feedback?
Model name, thread length, the reply that broke, and one line on what it used to do. See also Why we fine-tune Our4o ourselves.
Giovanni Gigantino
Founder of Get4oBack
