TMTPOST -- OpenAI CEO Sam Altman has revealed that the company is preparing to release its first open-weight language model with advanced reasoning capabilities since GPT-2 in 2019. "We've been thinking about this for a long time, but other priorities took precedence," Altman wrote in a post on X. "Now it feels important to do." He added that OpenAI is still finalizing key decisions and will host developer events to gather feedback before rolling out early prototypes. OpenAI has not yet provided additional details or responded to requests for comment. Open-weight AI models strike a balance between open-source and proprietary systems by sharing pre-trained parameters while keeping key development details, such as training code, datasets, and model architecture, undisclosed. This allows developers to use the model for inference and fine-tuning without full access to its underlying framework.