OpenAI has officially retired the o3 model family from ChatGPT, ending a 90-day sunset period that began with the May 28 announcement.
The o3 model, which was known for its strong reasoning capabilities and was a favorite among many power users, has been removed from the standard ChatGPT model selector and custom GPT options. Users who relied on o3 for complex reasoning tasks are now being directed to GPT-5 series models.
The retirement applies only to ChatGPT, not the API. Developers using o3 through the OpenAI API can continue to access the model, though it remains on a deprecation path.
OpenAI has been consolidating its model lineup throughout 2026, retiring older models as newer, more capable versions become available. The company retired GPT-4o, GPT-4.1, and GPT-4.1 mini earlier this year, along with GPT-5 (Instant, Thinking, and Pro) models.
The o3 retirement follows OpenAI's broader strategy of simplifying its product offerings and focusing resources on its most advanced models. The company has also retired the DALL·E GPT today (August 30) and plans to shut down Sora 2 on September 24.
Some users have expressed frustration on social media about the loss of o3, particularly those who valued its specific reasoning style. However, OpenAI maintains that the newer GPT-5 series models offer equivalent or better performance across all tasks.
