OpenAI said it is preparing for a next-generation open-source project
OpenAI recently hinted at the possibility of releasing the latest small artificial intelligence (AI) models as open-source. This announcement comes in light of China's DeepSeek launching an AI model as open source and Elon Musk-led xAI unveiling its model in the same manner. OpenAI, known for operating AI models in a closed manner, is now also considering a shift towards open-source initiatives.
CEO of OpenAI, Sam Altman, shared during a recent event that the next open-source project by OpenAI will focus on creating a model that is as small as the o3-mini but still requires a graphics processing unit (GPU) for optimal performance. The goal is to develop a model that can be highly utilized or one that can be run on a mobile phone.
OpenAI's Transition to Open Source
OpenAI introduced the o3-mini, a small model specialized for inference, in January. While the organization had previously kept its core model, the Large Language Model (LLM), closed, the unveiling of DeepSeek-R1 by China's DeepSeek as an open-source model prompted OpenAI to reconsider its strategic approach.
CEO Altman acknowledged a shift in perspective during a Reddit-hosted event, stating, "I personally think we have been on the wrong side of history (in relation to open-source software) and need to create other open-source strategies." He also mentioned during a meeting with developers in Korea that while OpenAI is exploring open-source strategies, there are no concrete plans in place yet.
Competing Open-Source AI Companies
OpenAI finds itself in the company of other open-source AI pioneers like Meta and DeepSeek. Elon Musk's xAI, a competitor to OpenAI, announced its new model, "Grook 3," and plans to release its previous version, "Grook 2," as open source in the coming months.
