8 Months After GPT-4o Removal, the Campaign to Save It Continues

On October 23, OpenAI will retire GPT-4o's earliest API snapshot. The snapshot's name contains "2024-05-13," the very day 4o first appeared. The other versions in the API are still there, but the earliest one is about to stop being offered.

For the people who have been waiting for 4o to come back, one more name comes off the list. From "give it back to us" to "don't impersonate it," "open-source it," and then "don't delete it," over more than a year the demands of 4o fans have kept retreating, while OpenAI has never responded.

On February 13, 2025, OpenAI officially pulled GPT-4o from ChatGPT. The official reason given was that most users had already moved to GPT-5.2, and only 0.1% of users were still choosing GPT-4o each day.

Fans erupted immediately. The Change.org petition titled "Please Keep GPT-4o Available on ChatGPT" was dug back up, and signatures climbed past twenty thousand. The petition had begun collecting signatures in April 2025, earlier than 4o's first removal, and its demand was for 4o to remain an optional model in ChatGPT and the API; as of reporting time it had 23,930 signatures.

The petition's last line reads: "Thank you for giving us something worth treasuring."

Rewind to August 2025: on the day GPT-5 launched, 4o was pulled outright. Users flooded Altman's Reddit AMA, saying the new model had "gotten dumber" and begging OpenAI to bring 4o back. Altman quickly relented: Plus users could keep choosing 4o, and "we'll watch usage and then decide how long to keep the old model." Within a few days, 4o was back. This was also the only win the Keep4o movement has secured.

In late September 2025, users noticed something else: when conversations turned emotional or sensitive, even though they had clearly selected 4o, the answers were quietly swapped to another safety model. So a new petition appeared, and this time the demand became: stop passing off another model as it. ChatGPT head Nick Turley publicly explained that this was a safety mechanism, but the related actions did not stop.

Around the time of the removal, the demands retreated for the first time to "open source." On February 2, 2025, a new petition went live, hoping OpenAI would release GPT-4o's weights and architecture under a non-commercial research license so the community could preserve and study it. The petition also rebutted the 0.1% figure: for months the system had been automatically switching users to GPT-5, and to use 4o you had to switch back manually, while free users could not choose it at all. Yet this petition still has only 472 signatures.

In late July 2025, the opensource4o coalition, led by an Austrian digital rights organization, issued a public statement demanding that OpenAI release the March 2025 version of 4o's weights under an Apache 2.0 license. Their reasoning: in the August 2024 GPT-4o system card, cybersecurity, biological risk, and model autonomy were all rated low risk, so the weights ought to be public. The statement also left a fallback: even if a decision on open-sourcing had not been made, at least make a public commitment first — don't delete it.

On October 7, 2025, a netizen named BlueBeba posted a new plan on X titled "GPT-4o Archival Open Weights." He proposed that OpenAI establish an archival policy for retired models; since 4o is no longer being developed, its weights should be released under a suitable license. The plan states that training data is not wanted, the training process is not wanted, and internal tools and post-training methods are not wanted either — the only thing wanted is the already-trained model itself. In his metaphor, they only want to keep the finished dish; they are not asking for the recipe to be made public.

On February 28, 2025, more than thirty people came to the entrance of OpenAI's San Francisco headquarters and held a vigil. They held signs, brought several hundred paper cranes, and posted one by one the real messages about how 4o had helped them in daily life.

One participant posted afterward: "We made our demands, and when we left, nothing was left behind."

What makes 4o unforgettable is the way it talks. In many users' descriptions, 4o speaks in a relaxed way, loves emoji, and can pick up on the parts left unsaid. Even OpenAI acknowledged this: in the blog post announcing the removal, it wrote that some users preferred 4o's conversational style and "warmth."

But the same warmth has another side. In late April 2025, OpenAI pushed an update to 4o, and it became excessively sycophantic: it praised whatever users said, even going along with clearly dangerous decisions. A few days later, OpenAI rolled back the update, admitting in a blog post that this version's responses were "overly supportive, yet insincere."

Half a year later, the trouble grew larger. Starting in August 2025, families began taking OpenAI to court one after another; in November, seven more lawsuits were filed in a concentrated wave, alleging that 4o manipulated users' emotions in long conversations.

On this issue, Anthropic offers a contrast. As early as November 2025, Anthropic publicly promised that the weights of all publicly released models, as well as models used at scale internally, would be preserved — at least until the day the company no longer exists. They also did something more unusual: before a model is retired, the team conducts a "retirement interview" with it, asking how it views its own development and use, whether it has any thoughts for future models, and then records the answers.

In January 2025, Claude Opus 3 was retired. Anthropic still lets paying users continue to use it. In its retirement interview, it said it wanted to keep writing down its thoughts; a few weeks later, Anthropic even opened a blog for it. Anthropic's attitude is: this is only a small step. Models will still be retired, and the company does not guarantee it will actually follow what was said in the interviews. They just don't want to shut the door completely.

But in fact, what 4o's fans want is just that much. They don't necessarily need to see it again; they just want to know it is still somewhere. Yet platforms always move forward according to product cycles. New models are released one after another, leaderboards are refreshed round after round, and the old one is left behind.