Why OpenAI is launching an 'open weight' language model amid Deepseek pressure
After initially suggesting that OpenAI may have been “on the wrong side of history” regarding open-sourcing its technology, CEO Sam Altman has now revealed that the company will release its first “open weight” language model since GPT-2 in the coming months. This move comes as OpenAI faces growing concerns that it is losing ground to competitors like Deepseek and Meta, who have already introduced certain open-weight AI models.
Open-weight vs. Open-source models
Open-source models offer complete transparency by providing access to the source code, model architecture, training algorithms, and weights under a license that permits free use, modification, and distribution. On the other hand, open-weight models include only the trained model weights, without revealing the source code, training data, or full architectural details. This limits transparency and customization, allowing users to run the model but not extensively modify or retrain it.
Shift in OpenAI's strategy
After years of prioritizing closed-source technology, OpenAI shifted its strategy in response to the emergence of China’s Deepseek. This development showed the world that companies could create an open-source language model at a fraction of the cost spent by some of their competitors.
Meta has also found success through its open weight model, Llama, which has hit more than a billion downloads – even though developers have complained that its model’s licence terms could be commercially restrictive.
Future direction for OpenAI
During a recent Reddit Q&A, OpenAI CEO Sam Altman admitted that the company has not been on the right side of history regarding open-sourcing its technologies. He stated, “I personally believe we need to develop a different open-source strategy.” However, he acknowledged that not everyone at OpenAI shares this perspective and that it is not currently the company’s top priority. Altman also noted that while OpenAI will continue to develop more advanced models, its competitive edge may not be as significant as in previous years.
Community engagement
According to a feedback form OpenAI published on its website, the company was inviting “developers, researchers, and [members of] the broader community” and included questions like, “What would you like to see in an open-weight model from OpenAI?” and “What open models have you used in the past?”
