OpenAI Seeks to Make its upcoming Open AI Model Best-in-Class ...
Toward the end of March, OpenAI announced its intention to release its first “Open” language model assembly GPT-2. Now, details about that model are beginning to trickle out from the company’s sources.
Development of the Open Model
Sources tell TechCrunch that Aidan Clark, OpenAI’s VP of Research, is leading the development of the Open Model, which is in the very early stages. OpenAI is targeting an early summer release and aims to model the lines of OpenAI’s current benchmark-topping reasoning models.
Exploring a Permissive License
OpenAI is exploring a highly permissive license for the model with minimal restrictions, according to TechCrunch’s sources. This is in contrast to models like Llama and Google’s Gemma, which have faced criticism for imposing requirements.
Increasing Pressure from Rivals
OpenAI is facing increasing pressure from rivals such as the Chinese AI lab Deepseek, which have adopted a more open approach to launching models. These "open" competitors make their models available to the AI community for experimentation and, in some cases, commercialization.
Focus on Safety and Evaluation
OpenAI intends for its Open Model, which will be "Text in, Text Out," to run on high-end consumer hardware and possibly allow developers to switch it for anthropic and other purposes. CEO Sam Altman emphasized the importance of thorough safety evaluations for the upcoming model.
Model Card Release
The company intends to release a Model Card for the Open Model before the official release to ensure transparency and safety. Altman stated that the model will undergo rigorous evaluation according to their preparedness framework.
Continuing Discussions on Open Source Strategy
Altman acknowledged the need for a different open-source strategy going forward and emphasized the importance of maintaining a balance between innovation and accessibility. OpenAI aims to produce better models while keeping the community engaged and informed.
OpenAI has faced criticism for rushing safety testing of recent models and failing to release Model Cards for others. Altman remains committed to addressing these concerns and ensuring that the upcoming Open Model sets a new standard for transparency and accountability.
