OpenAI delays launch of its open-weight AI model: All you need to know

OpenAI delays release of its open-weight AI model to later this summer, as CEO Sam Altman said a major research breakthrough could enhance its performance and features

OpenAI has delayed the release of its open-weight model, which was initially expected to be released in early summer. CEO Sam Altman announced the delay through a post on X (formerly Twitter) on June 11, stating “we are going to take a little more time with our open-weights model, i.e. expect it later this summer but not June."

 

Altman shared that the delay comes as a result of unexpected progress made by OpenAI’s research team. He explained “Our research team did something unexpected and quite amazing, and we think it will be very, very worth the wait, but needs a bit longer.” 

OpenAI’s last open model was GPT-2, which was released in 2019. Since then, the AI start-up company has released several closed models including GPT-4.1, o3 series, o4 series and more. 

OpenAI’s open-weight model: What to expect

The upcoming open model is said to feature reasoning capabilities on par with OpenAI’s o-series models and aims to outperform other leading open reasoning models like DeepSeek’s R1.

 

OpenAI is exploring advanced features to strengthen its open model. According to a report by TechCrunch, one possibility includes integrating the model with OpenAI’s cloud-based systems for handling complex tasks. However, it remains uncertain whether such capabilities will be part of the final release.

What are open-weight language models?

Open-weight language models are AI systems whose trained parameters—the numerical values that guide how the model operates—are openly shared with the public. This enables researchers, developers, and organisations to download and use the models on their own devices, without needing access to a cloud-based API (Application Programming Interface). However, such models can still come with licensing limitations, especially around modifications and commercial applications. 

How are they different from other models?

Open-weight models strike a balance between fully open-source and entirely closed AI systems. Unlike open-source models that offer complete access to both code and model weights with minimal restrictions, open-weight models typically release only the trained weights and may include limitations on how they can be modified or used commercially. On the other hand, closed models like OpenAI’s GPT-4 and Google’s Gemini keep both their code and weights private, maintaining full proprietary control over their technology.