OpenAI is preparing to release an open-weight large language model as early as next week, according to people familiar with the plans. The forthcoming system, described as comparable in size and reasoning performance to the "o3 mini" model, would be the company’s first openly available set of model weights since it published GPT-2 in 2019. OpenAI intends to distribute the model through Microsoft Azure, Hugging Face and other cloud providers, giving developers and enterprises the ability to run it on their own infrastructure. The move broadens access to OpenAI technology and could reshape the firm’s relationship with Microsoft, which secured exclusive sales rights for most OpenAI models under a 2023 pact that also gives each side a 20 percent share of the other’s AI revenues. By allowing rival cloud operators to host the software, the open-weight release may undercut Azure’s pricing power and intensify competition among cloud platforms just as the partners renegotiate their commercial arrangements.