OpenAI Delays Once Again Launch of Its Open Model

On Friday, OpenAI CEO Sam Altman announced the company will delay its open model release, marking another one-month postponement. Originally scheduled for early summer, the model's launch has now been indefinitely postponed to conduct additional safety testing.

"We need time for extra safety evaluations and to review high-risk areas. We're uncertain about the exact timeline," Altman posted on X. "While we believe the community will create remarkable applications using this model, once the weights are released, there's no taking them back. This presents new challenges for us, and we're committed to doing it right."

The open model launch from OpenAI was one of the most anticipated AI events this summer, alongside the expected GPT-5 release from ChatGPT's creators. Unlike GPT-5, OpenAI's open model would allow developers to freely download and run locally. Through these dual launches, OpenAI aims to demonstrate its position as Silicon Valley's leading AI laboratory - a challenge growing more difficult as xAI, Google DeepMind, and Anthropic invest billions into their own projects.

This delay means developers must wait longer to access OpenAI's first open model in years. Previously reported by TechCrunch, OpenAI's open model is expected to exhibit reasoning capabilities comparable to the company's o-series models and will be positioned as the best-in-class offering.

This week intensified competition within the open AI model ecosystem. Earlier Friday, Chinese AI startup Moonshot AI unveiled Kimi K2, a trillion-parameter open AI model outperforming OpenAI's GPT-4.1 on multiple agent coding benchmarks.

When Altman first announced the initial delay in June, he mentioned achieving "unexpected and quite impressive results" without providing specifics.

"In terms of capabilities, we consider this model exceptional - but our standards for open-source models are high," stated Aidan Clark, OpenAI's research vice president leading open model initiatives in a X post. "We believe more time is needed to ensure we deliver a model we can truly be proud of across all aspects."