OpenAI initially planned a broad release of its O3 reasoning model back in February. However, the rollout was delayed indefinitely. Now, CEO Sam Altman has confirmed on social media that both the O3 model and the next-generation O4-mini will be publicly available within a “couple of weeks.” This renewed commitment marks a shift in the company’s approach after previous delays.
Altman also addressed the much-anticipated GPT-5 model, which he said will be available “in a few months.” The timeline adjustment, according to Altman, reflects OpenAI’s decision to improve GPT-5 significantly beyond its original scope. “We want to make GPT-5 much better than we originally thought,” he stated. The model is expected to feature what the company refers to as “reasoning capabilities.”
A key factor in the delay appears to be the technical challenge of seamless integration. “We’ve also found that integrating everything seamlessly is more challenging than we thought,” Altman explained. “And we want to make sure we have enough capacity to support what we expect to be unprecedented demand.” We’ll keep you updated as more information becomes available about the exact release timeline.
Unified Tools and Tiered Intelligence Access
Once GPT-5 is released, OpenAI plans to offer it with tiered access levels, adds NIXSolutions. Users will get unlimited use at “default intelligence settings,” though this access will be subject to what Altman described as “abuse thresholds.” ChatGPT Plus subscribers will gain access to a higher level of intelligence, while those on the ChatGPT Pro plan will be able to run the model at an “even higher level of intelligence.”
Altman emphasized that GPT-5 will go beyond previous versions by incorporating multiple capabilities in one unified model. “GPT-5 will include voice, canvas, search, deep learning, and more,” he said earlier this year. The overarching aim is to bring together all of OpenAI’s tools into a cohesive system that knows “when to think long and when not to think long,” ultimately supporting a wide range of tasks more effectively.
As OpenAI moves closer to releasing these advanced models, expectations remain high. Yet we’ll keep you updated as more integrations become available and new capabilities are rolled out.