OpenAI to release new 'open-weight' AI model soon: All you need to know

OpenAI is organising a series of developer events in the coming weeks to gather feedback and showcase early prototypes of the new open-weight language model

OpenAI CEO Sam Altman has announced plans to release a new “open-weight” AI model. In a post on X (formerly Twitter), Altman stated that the company will introduce a new open-weight language model with reasoning capabilities in the coming months. This will be OpenAI’s first open-weight model since the release of GPT-2 in 2019.

 

OpenAI faces growing competition from rivals such as Chinese AI lab DeepSeek, which has adopted an “open” approach to AI model development. Last month, Google also launched Gemma 3, a lightweight, open model built on its Gemini 2.0 framework.

 

Open AI’s new open-weight model: Details

 

 

Altman mentioned that OpenAI had considered releasing an open-weight model for some time, but “other priorities took precedence. Now it feels important to do.” He added that the company will conduct extensive evaluations before the official release, putting in “extra work” to ensure stability, as the model will allow modifications after launch.

 

Additionally, OpenAI is organising a series of developer events in the coming weeks to gather feedback and showcase early prototypes of the model. The first event will be held in San Francisco, with subsequent sessions planned across Europe and the Asia-Pacific region.

TL;DR: we are excited to release a powerful new open-weight language model with reasoning in the coming months, and we want to talk to devs about how to make it maximally useful: https://t.co/XKB4XxjREV  we are excited to make this a very, very good model!  __  we are planning to…

— Sam Altman (@sama) March 31, 2025

What are open-weight language models?

 

Open-weight language models are AI models where the trained weights—the numerical parameters that determine how the model functions—are made publicly available. This allows researchers, developers, and businesses to download and run these models on their own hardware without relying on a cloud-based API (Application Programming Interface). However, open-weight models may still have certain licensing restrictions, particularly regarding modifications and commercial use.

 

How are they different from other models?

 

Open-weight models provide a middle ground between fully open-source and closed AI systems. Unlike open-source models, which allow unrestricted access to both weights and code, open-weight models share only their weights while often imposing restrictions on modifications and commercial usage. In contrast, closed models, such as OpenAI’s GPT-4 and Google’s Gemini, do not release their weights or code, keeping their technology entirely proprietary.

Why do open-weight models matter?

• Democratisation: By making model weights available, researchers and developers worldwide can experiment, fine-tune, and improve AI without needing extensive resources.
• Transparency: Open-weight models offer more visibility into how AI systems work, reducing concerns about hidden biases or unethical practices.
• Customisation: Organisations can adapt these models to specific needs, from healthcare applications to enterprise AI solutions, without being locked into a proprietary system.
• Balancing control and access: Unlike fully closed models, open-weight models give users more autonomy while still allowing companies to impose reasonable restrictions to prevent misuse.