OpenAI Plans to Release a New 'Open' AI Model in the Coming ...
Despite its name, OpenAI can’t claim to be completely open. Recent language models from the ChatGPT team have been closed and proprietary, drawing criticism from various quarters, including one of its founders, Elon Musk.
However, there seems to be a shift in the works. The current CEO of OpenAI, Sam Altman, has revealed plans to release a new open-weighted language model in the coming months. This model will be capable of running locally, a departure from the current reliance on the company’s servers. The last model with similar features released by OpenAI was GPT-2, which came out six years ago.
Embracing Openness
Announced through a publication on X, the decision marks a significant shift in OpenAI's approach. Altman expressed, "We’re planning to release our first open weights language model since GPT-2. We’ve been thinking about this for a long time, but other priorities took precedence. Now we feel it’s important to do so."
Unlike open-source models, where developers have access to source code, training data, and complete architecture, an open-weight model offers more limited openness. Users will have the ability to fine-tune the project but won't be able to rebuild it from scratch due to restrictions on access to critical components like the original dataset or full architectural details.
Competitive Landscape
The move comes at a time of stiff competition in the artificial intelligence sector, with players like Meta, Google, and DeepSeek introducing their own advanced models. Altman's decision to push for an open-weight model has not been without internal debate at OpenAI, as revealed in Decrypt's coverage of the company.
Despite dissenting voices, Altman emphasized the need to reconnect with the AI community and adapt to the evolving competitive environment. OpenAI is actively seeking developer feedback through a form on its website and plans to host in-person events in multiple regions.
Looking Ahead
As details about the new model remain scarce, questions linger about its parameters, context window size, training methods, and licensing terms. Altman hinted that the model's reasoning capabilities would rival o3-mini, positioning it as a leading reasoning model among open-weight variants.
While the launch of the new model isn't an immediate priority for OpenAI, the company recently closed a funding round valuing it at $300 billion. With input from researchers and developers, OpenAI intends to finalize the features of the model and release it "in the coming months." The move aligns with the company's commitment to innovation while navigating a rapidly evolving AI landscape.
