OpenAI to retire GPT-4o's May 2024 API snapshot on October 23

OpenAI will discontinue gpt-4o-2024-05-13, the model's earliest API snapshot, on October 23, according to QbitAI. Later GPT-4o API versions remain available, making this a snapshot retirement rather than a complete API withdrawal. The deadline follows GPT-4o's February 13 removal from ChatGPT and renewed pressure from Keep4o supporters. An October 7 proposal asks OpenAI to release the retired model's weights under an appropriate license without disclosing training data, development processes, or internal tools. The report describes no OpenAI commitment to do so.

The dispute highlights a foundation-model market tension: labs advance their product portfolios, while some customers value a particular model's conversational behavior over newer capabilities. OpenAI acknowledged users' preference for GPT-4o's style and warmth, but said only 0.1% of users selected it daily when explaining its ChatGPT retirement. That figure describes selection within ChatGPT, not API demand. Preservation requests also expose three distinct questions: whether a lab retains weights internally, continues serving them, or releases them for others to operate. None guarantees the others, and user petitions establish demand for continuity without demonstrating its commercial scale.

Builders using this exact snapshot should identify affected integrations and test replacements against their own workflows before October 23, including conversational tone and multimodal behavior. Later GPT-4o snapshots provide another migration option, but their continued availability should not be treated as a permanent guarantee. For investors, the concrete question is whether model providers can manage retirement without disrupting applications or weakening customer trust; the source establishes neither the cost of continued service nor the feasibility of releasing the weights.