OpenAI is preparing to release its first truly open language model since GPT-2, according to sources familiar with the company’s plans. The new AI model, focused on reasoning capabilities, is expected to launch in early summer under a highly permissive license with minimal usage restrictions.
OpenAI to Release Open-Source AI Model
Aidan Clark, OpenAI’s VP of research, is spearheading the development of this model, which remains in its early stages. The company aims to create a “text in, text out” system that can run on high-end consumer hardware and potentially allow developers to toggle its reasoning capabilities on or off – similar to features in recent models from competitors like Anthropic.
This move represents a significant shift for OpenAI, which has faced growing competition from rivals embracing more open approaches to AI development.
Chinese AI lab DeepSeek and Meta have gained substantial traction with their open models, with Meta’s Llama family reportedly surpassing 1 billion downloads as of March.
CEO Sam Altman has previously acknowledged that OpenAI may have been “on the wrong side of history” regarding open-sourcing its technologies.
During a January Reddit Q&A, Altman stated: “I personally think we need to figure out a different open source strategy. Not everyone at OpenAI shares this view, and it’s also not our current highest priority.”
The decision to release an open model comes as OpenAI navigates increasing pressure from both competitors and the AI community. While companies like Meta and Google have released open models, they’ve faced criticism for imposing what some consider excessive restrictions on usage and commercialization – a pitfall OpenAI apparently hopes to avoid with its more permissive approach.
Safety remains a key consideration for the upcoming release. Altman has promised thorough safety evaluations, stating on X (formerly Twitter) last month: “Before release, we will evaluate this model according to our preparedness framework, like we would for any other model, and we will do extra work given that we know this model will be modified post-release.”
OpenAI to Release Detailed Model Card Amid Safety Scrutiny
Sources indicate OpenAI plans to publish a detailed model card documenting its internal and external benchmarking and safety testing results.
This emphasis on safety comes amid criticism from AI ethicists who have accused OpenAI of rushing safety testing for previous models and failing to release comprehensive model cards. Altman himself faced allegations of misleading OpenAI executives about model safety reviews prior to his brief removal as CEO in November 2023.
If successful, this summer release could be just the beginning of OpenAI’s new open strategy. Sources suggest the company may follow up with additional models, including potentially smaller versions, depending on the reception of this initial offering.
The upcoming model represents OpenAI’s attempt to balance competitiveness in an increasingly crowded AI landscape with responsible development practices.
As Altman noted earlier this year, while OpenAI expects to “produce better models going forward,” the company anticipates maintaining “less of a lead than we did in previous years” – an acknowledgment of the rapidly evolving and increasingly open AI development ecosystem.
OpenAI had not responded to requests for comment at the time of publication.
