OpenAI has announced a delay in the release of its first open AI model in years. Originally expected in June, the model will now launch later this summer, according to CEO Sam Altman’s post on X (formerly Twitter).
“We are going to take a little more time with our open-weights model,” Altman said. “Our research team did something unexpected and quite amazing. We think it will be very worth the wait, but it needs a bit longer.”
The upcoming open model is expected to have strong reasoning capabilities, comparable to OpenAI’s proprietary o-series models. It is designed to outperform other open-source reasoning models, such as DeepSeek’s R1.
Since OpenAI announced its plans, competition in the open-source AI space has intensified. Mistral, known for releasing open models, introduced its new Magistral reasoning model family on the same day as Altman’s post. In April, Chinese lab Qwen also launched hybrid reasoning models capable of switching between in-depth reasoning and fast responses.
To remain competitive, OpenAI has reportedly considered adding advanced features to its open model. These may include the ability to connect with OpenAI’s cloud-hosted models for handling more complex tasks. However, it’s not yet clear if those capabilities will be included in the final release.
The release of this open model is seen as a key move for OpenAI to rebuild its relationship with developers and the broader research community. Altman has previously acknowledged that OpenAI has fallen behind in openness, admitting the company was on the “wrong side of history” regarding open-source contributions.
By delivering a powerful open model, OpenAI hopes to re-establish itself as a leader in the open AI ecosystem and meet growing expectations in a rapidly evolving market.
