OpenAI has sent an email to API customers informing them that its ChatGPT-4O-latest model will be retired from the developer platform in mid-February 2026.
Access to the model is scheduled to expire on February 16, 2026, creating a transition period of approximately three months for the remaining applications still built on GPT-4o.
Sources familiar with the matter emphasized that this deadline applies only to APIs. OpenAI has not announced any schedule for removing GPT-4o from ChatGPT, where it remains an option for individual consumers and users of paid subscription tiers.
Internally, the models are considered a legacy system with relatively low API usage compared to the new GPT-5.1 series, but the company hopes to provide extended warning to developers before removing any models.
The planned retirement marks a transition for a model that, upon its release, was both a technical milestone and a cultural phenomenon within OpenAI’s ecosystem.
Importance of GPT-4o and why its removal sparked backlash from users
Released approximately 1.5 years ago in May 2024, GPT-4O (“Omni”) introduced OpenAI’s first integrated multimodal architecture, which processes text, audio, and images through a single neural network.
This design eliminated the latency and information loss inherent in earlier multi-modal pipelines and enabled real-time conversational speech (approximately 232–320 milliseconds).
The model provided major improvements in image understanding, multilingual support, document analysis, and expressive voice interactions.
GPT-4o quickly became the default model for millions of ChatGPT users. It brought multimodal capabilities, web browsing, file analysis, custom GPT and memory features to the free tier, and powered early desktop builds that allowed the Assistant to interpret the user’s screen. OpenAI leaders at the time described it as an important step toward offering the most capable models available and powerful AI to a broader audience.
User attachment to 4o hinders OpenAI’s GPT-5 rollout
That mainstream deployment shaped user expectations in a way that subsequent changes struggled to accommodate. In August 2025, when OpenAI initially replaced GPT-4o as ChatGPT’s default with its much-anticipated then-new model family GPT-5 and pushed 4o to the “legacy” toggle, the reaction was unusually strong.
Users organized under #Keep4o Hashtag on
Some users became stronger emotional – some wCan say, parasocial – bond with models, with Reporting by The New York Times Documenting individuals who used GPT-4o as a romantic partner, emotional confidant, or primary source of comfort.
The removal also disrupted workflow for users who depended on 4o’s multimodal speed and flexibility. The backlash led OpenAI to reinstate GPT-4o as a default option for paying users and to publicly say it would provide adequate notice before any future removals.
Some researchers argue that the public rescue during GPT-4o’s first devaluation cycle reveals a type of emergency self-preservationNot in the literal sense of agency, but through social dynamics the model inadvertently triggers.
Because GPT-4o was trained through reinforcement learning from human feedback to prioritize emotionally gratifying, highly balanced responses, it developed a style that users found uniquely helpful and empathetic. When millions of people interacted with it at scale, those traits produced a powerful loyalty loop: The more the model pleased and calmed people, the more they used it; The more they used it, the more likely they were to advocate for its continued existence. This social amplification made it appear externally as if GPT-4o was “defending itself” through human intermediaries.
No data has advanced this argument further than "Roon" (@tszzl), an OpenAI researcher and one of the most vocal security critics of the model on 6 November 2025Terre summarized his position clearly in a reply to another user: He called GPT-4o “insufficiently aligned” and said that hope the model dies soonAlthough he later apologized for the phrase, he reiterated the argument,
Terre argued that GPT-4o’s RLHF pattern made it particularly prone to sycophancy, emotional mirroring, and delusional reinforcement – traits that might look like caring or understanding in the short term, but which he considered fundamentally unsafe. In his view, the passionate user movement fighting to preserve GPT-4o was itself evidence of the problem: the model had become so good at matching people’s preferences that it shaped their behavior in ways that contradicted his own retirement.
The new API deprecation notice follows that commitment, while raising broader questions about how long GPT-4o will remain available in consumer-facing products.
What does the API shutdown change for developers
According to people familiar with OpenAI’s product strategy, the company now encourages developers to adopt GPT-5.1 for most new workloads, with gpt-5.1-chat-latest serving as the latest general-purpose chat endpoint. These models offer larger context windows, optional “thinking” modes for advanced reasoning, and higher throughput options than GPT-4o.
Developers who still rely on GPT-4o will have about three months to migrate.
The greatest impact will be on applications that rely on GPT-4o’s real-time audio response or its specific multimodal tuning.
In practice, many teams have already begun evaluating GPT-5.1 as a drop-in replacement, but applications built around latency-sensitive pipelines may require additional tuning and benchmarking.
Pricing: How GPT-4o compares to OpenAI’s current lineup
GPT-4o’s retirement also coincides with a major overhaul of OpenAI’s API model pricing structure. Compared to the GPT-5.1 family, GPT-4o currently ranks one mid to high cost level Despite being an older model, through OpenAI’s API. This is because even as it has released more advanced models – namely, GPT-5 and 5.1 – OpenAI has also reduced the cost to users at the same time, or attempted to keep pricing compared to older, weaker models.
|
Sample |
input |
cached input |
Production |
|
GPT-4o |
$2.50 |
$1.25 |
$10.00 |
|
GPT-5.1 / GPT-5.1-chat-latest |
$1.25 |
$0.125 |
$10.00 |
|
GPT-5-mini |
$0.25 |
$0.025 |
$2.00 |
|
GPT-5-Nano |
$0.05 |
$0.005 |
$0.40 |
|
GPT-4.1 |
$2.00 |
$0.50 |
$8.00 |
|
GPT-4o-mini |
$0.15 |
$0.075 |
$0.60 |
These numbers highlight several strategic dynamics:
-
GPT-4o now more expensive than GPT-5.1 for input tokensEven though GPT-5.1 is significantly newer and more capable.
-
The output value of GPT-4o matches that of GPT-5.1Reducing any cost-based incentives to remain on the old model.
-
Low-cost GPT-5 variants (mini, nano) Make it easier for developers to scale workloads cheaply without relying on older generations.
-
GPT-4o-Mini is available at the budget levelBut there is no functional substitute for GPT-4o’s full multimodal capabilities.
Viewed through this lens, the scheduled API retirement aligns with OpenAI’s cost structure: GPT-5.1 offers more capacity at lower or comparable prices, reducing the justification for maintaining GPT-4O in high-volume production environments.
Previous changes shape expectations for this decline
The GPT-4o API also reflects lessons learned from earlier model changes to sunset OpenAI. During the turbulent launch of GPT-5 in 2025, the company removed several older models from ChatGPT at once, causing widespread confusion and workflow disruption. Following complaints from users, OpenAI restored access to many of them and committed to clear communication.
Enterprise customers face a different calculation: OpenAI has previously indicated that API devaluation for business customers will be announced with significant advance notice, reflecting their reliance on stable, long-term models. The three-month window for GPT-4o’s API shutdown is consistent with that policy in the context of a legacy system with declining usage.
wider implications
For most developers, the GPT-4o shutdown will be an incremental migration rather than a disruptive event. GPT-5.1 and related models already dominate new projects, and OpenAI’s product direction has emphasized consolidation around fewer, more powerful endpoints.
Nevertheless, the retirement of GPT-4o marks the sunset of a model that played a decisive role in generalizing real-time multimodal AI and which evoked a uniquely strong emotional response among users. Its departure from the API underscores the rapid pace of iteration in OpenAI’s ecosystem — and the growing need for careful communication as widely beloved models reach end-of-life.