OpenAI’s CEO has once again announced a delay in the company’s major project.
Sam Altman, CEO of OpenAI, stated that the release of the company’s open-weight model—already delayed once—has been postponed again, this time without a specific timeline for publication. Altman explained that his team needs more time to conduct safety tests and assess high-risk components.
In a post on X, Altman wrote: “As long as the model weights are not released, we can prevent potential mistakes; but once they’re out, there’s no going back. This is a new experience for us, and we want to do it right.”
The release of OpenAI’s open model, alongside the potential launch of GPT-5, was considered one of the most significant AI events of the summer. Unlike GPT-5, the open-weight model is expected to be available for download and local execution by developers.
With the release of new models, OpenAI aims to maintain its position as the leading AI company in Silicon Valley—though this goal is becoming increasingly challenging amid heavy investments by competitors such as xAI, DeepMind, and Anthropic.
It is said that OpenAI’s open-weight model, the creator of ChatGPT, will be comparable in reasoning capabilities to the company’s “O” series models, and is expected to outperform all existing open-source models in terms of quality.
Competition in the AI field has intensified. Earlier today, Chinese company Moonshot AI introduced the Kimi K2 model, which has one trillion parameters and has outperformed GPT-4.1 in several agent-based coding benchmarks.
