OpenAI to Release New Open-Source AI Language Model

OpenAI's Forthcoming Open Language Model

OpenAI has announced its intention to release its first openly accessible language model since the introduction of GPT‑2, with availability expected “in the coming months.”

Gathering Community Input

This information stems from a feedback form recently published on the company’s website. The form is directed towards “developers, researchers, and the broader community” and seeks insights into desired features. Questions include inquiries about preferred functionalities in an open-weight model and past experiences with open models.

OpenAI expressed its enthusiasm for collaboration, stating, “We’re excited to collaborate with developers, researchers, and the broader community to gather inputs and make this model as useful as possible.” Interested parties are invited to participate in feedback sessions via the provided form.

Developer Engagement and Events

The company is planning a series of developer events to collect feedback and showcase model prototypes. The initial event is scheduled to be held in San Francisco within the next few weeks.

Further sessions will be organized in Europe and the Asia-Pacific region, broadening the scope of community involvement.

Responding to Competitive Pressure

OpenAI is experiencing increased competition from companies like DeepSeek, a Chinese AI lab, which have embraced an “open” model release strategy.

Unlike OpenAI’s previous approach, these competitors make their models readily available for experimentation and, in certain instances, commercial applications.

Success of Open Models

This strategy has proven remarkably successful for some organizations. Meta, with its substantial investment in the Llama family of open AI models, reported over 1 billion downloads of Llama earlier in March.

Similarly, DeepSeek has rapidly gained a significant global user base and attracted considerable investment from domestic sources.

Shift in OpenAI's Perspective

During a recent Reddit Q&A session, OpenAI CEO Sam Altman acknowledged that the company may have been mistaken in its reluctance to open-source its technologies.

Altman stated, “[I personally think we need to] figure out a different open source strategy.” He also noted that this viewpoint isn't universally shared within OpenAI and isn't currently the top priority.

He anticipates continued model improvements but expects OpenAI’s lead in the field to diminish in the coming years.

Details on the Upcoming Model

Expanding on the open model plans in a post on X, Altman revealed that the forthcoming model will possess “reasoning” capabilities comparable to OpenAI’s o3-mini.

“[B]efore release, we will evaluate this model according [to] our preparedness framework, like we would for any other model,” Altman explained. “[A]nd we will do extra work given that we know this model will be modified post-release.”

OpenAI is eager to observe how developers utilize the model and how larger organizations and governments implement it for self-hosted applications.

Allegations of Misleading Information

Excerpts from an upcoming book by Wall Street Journal reporter Keach Hagey suggest that Altman may have provided inaccurate information to OpenAI executives regarding model safety reviews prior to his temporary removal from his position in November 2023.