Originally launched last year as a flagship model, GPT-4o can infer audio, vision and text in real time. OpenAI will release GPT-4.1 as soon as next week, as well as smaller GPT-4.1 mini and nano versions.
OpenAI is also preparing a full version of its o3 inference model and an o4mini version that may be released earlier. AI engineer Tibo Blaho discovered references to o4mini, o4minihigh, and o3 in the new ChatGPT web version earlier today, suggesting these new additions are coming soon. Unless OpenAI adjusts its release schedule, both o3 and o4mini will be released next week.
OpenAI CEO Sam Altman revealed on X that OpenAI is launching an exciting feature today, but it's unclear if this is related to the o3 and o4mini models mentioned in ChatGPT. Sources cautioned that OpenAI has recently delayed the launch of some new models due to production capacity issues, so the launch of the new GPT-4.1 model may be delayed to next week's planned release.
Altman revealed at X earlier this month that customers "should expect new version launches of OpenAI to be delayed, issues may arise, and service to be slow at times due to the capacity challenges we face." OpenAI's more advanced image generation capabilities forced the company to temporarily throttle request rates last month, with Altman claiming "our GPUs were smoking" because the built-in image generator was too popular for ChatGPT free tier users.