Archive — US — techcrunch
OpenAI plans to release a new ‘open’ AI language model in the coming months
OpenAI plans to release its first 'open' language model since GPT-2 within months, seeking community feedback for development. This move, influenced by competing firms adopting open models, aims to enhance collaboration and accessibility. OpenAI intends to conduct developer events globally, emphasizing the new model's reasoning capabilities and its potential for independent deployment by users.
Key points
- OpenAI aims to release a new 'open' language model in the coming months, the first since GPT-2.
- The company is soliciting feedback from developers and researchers via a form on its website.
- Upcoming developer events will be held in San Francisco, Europe, and Asia-Pacific to gather insights.
- OpenAI's decision is partly driven by competition from other firms like DeepSeek and Meta that have adopted open model strategies.
- CEO Sam Altman acknowledged OpenAI's past mistakes regarding open sourcing technologies.
- The new model is expected to have advanced reasoning capabilities and will be evaluated before release.
Why it matters
- The trend towards open-source AI is gaining traction as seen with Meta's Llama model, which achieved significant downloads.
- Competition in the AI space, especially from companies like DeepSeek, is pressuring OpenAI to rethink its strategies.
- The conversation around AI fairness and accessibility is becoming increasingly critical as organizations move towards open models.
OpenAI's upcoming open language model reflects a strategic shift responding to competitive pressures, aiming to foster community engagement and innovation while addressing past criticisms of its closed-source approach.