OpenAI Delays Highly Anticipated Open Model Release Indefinitely for Enhanced Safety Testing

OpenAI, the leading artificial intelligence research company, has announced a further delay in the release of its highly anticipated open model. Originally slated for release next week, the launch has been pushed back indefinitely to allow for extensive additional safety testing and thorough review of high-risk areas.

This marks the second postponement for the open model this summer, a development that has sent ripples through the AI community. OpenAI CEO Sam Altman confirmed the delay in a post on X, stating, “We need time to run additional safety tests and review high-risk areas. we are not yet sure how long it will take us.” Altman emphasized the gravity of the decision, noting, “While we trust the community will build great things with this model, once weights are out, they can’t be pulled back. This is new for us and we want to get it right.”

The open model is one of the most keenly awaited AI events, paralleling the excitement around OpenAI’s upcoming GPT-5. Unlike its more widely known counterparts, OpenAI’s open model is designed for developers to download and run locally, offering unprecedented accessibility and flexibility. Its release is critical for OpenAI as it strives to maintain its position as Silicon Valley’s premier AI lab amidst escalating competition from powerhouses like xAI, Google DeepMind, and Anthropic, all of whom are investing billions into their own AI endeavors.

Developers will now have to wait longer to access what would be OpenAI’s first open model in years. Previous reports indicated that this model is expected to possess reasoning capabilities akin to the company’s ‘o-series’ models, with OpenAI aiming for it to be “best-in-class” compared to other open models on the market.

The competitive landscape for open AI models has intensified recently. Just prior to OpenAI’s announcement, Chinese AI startup Moonshot AI launched Kimi K2, a one-trillion-parameter open AI model that has reportedly outperformed OpenAI’s GPT-4.1 model in various agentic-coding benchmarks, underscoring the rapid advancements across the industry.

In a related statement regarding the delay, Aidan Clark, OpenAI’s VP of Research leading the open model team, posted on X: “Capability wise, we think the model is phenomenal — but our bar for an open source model is high and we think we need some more time to make sure we’re releasing a model we’re proud of along every axis.” This reiterates the company’s commitment to quality and safety over speed.

Earlier in June, when the initial delay was announced, Altman hinted at an “unexpected and quite amazing” achievement related to the model, without providing further specifics. Discussions have also reportedly included enabling the open AI model to connect to OpenAI’s cloud-hosted AI models for more complex queries, though it remains uncertain if this feature will be part of the final release.