According to a recent article on Fortune.com, OpenAI, co-founded by Sam Altman in 2015, has achieved significant milestones, including raising $40 billion at a $300 billion valuation, the largest funding round ever by a private tech company. The company’s new GPT-4o image generation model, producing Studio Ghibli-style images, has gone viral, and ChatGPT’s user base has expanded to 500 million weekly users, up from 400 million the previous month.
However, alongside these achievements, Altman announced a strategic shift for OpenAI with plans to release an open-source model in the coming months. This release would mark OpenAI’s first open model since GPT-2 in 2019, contrasting its recent trend towards closed models. Although not entirely open, the model will allow researchers and developers access to the underlying code and weights, enabling them to use, modify, or improve it.
The shift may be influenced by the rise of open-source initiatives, such as China’s DeepSeek, which altered the AI landscape favoring open-source practices. Industry insiders suggest that as AI technologies integrate into various business operations, there is a growing demand for the flexibility and transparency of open-source models. The narrowing performance gap between OpenAI and its competitors further pressures the company to adapt its strategy, a challenge Altman acknowledged earlier this year.
Naveen Rao, VP of AI at Databricks, noted that this move signifies a recognition of shifting value in AI from models themselves to their applications, allowing OpenAI to expand its presence in various use cases beyond ChatGPT. Rowan Curran, a senior analyst at Forrester Research, emphasized the growing importance of open-source AI as it provides flexibility for enterprises to operate models across different platforms and infrastructures.
The transition to open source is a complex balancing act for OpenAI. Previously, OpenAI’s concerns included protecting its models from foreign competitors, but as Chinese startups catch up, OpenAI risks losing influence if it does not engage with the open-source community. Chris Lehane, OpenAI’s Chief Global Affairs Officer, highlighted the need for a balanced approach between open and closed models to maintain strategic advantages and prevent misuse.
OpenAI’s open-source endeavor also appears to be competitive with Meta, which is releasing the fourth version of its open Llama model. Altman critiqued Meta’s licensing strategy on social media, indicating an intention to offer a more accessible open-source model without user restrictions.
Curran, however, remains skeptical about the openness of OpenAI’s forthcoming model, predicting it might still lack transparency compared to other open versions from rivals like Meta and Mistral.
