OpenAI Delays Release of Its Open AI Model—Again

Key Takeaways:

• OpenAI has delayed the release of its open AI model indefinitely due to unresolved safety concerns.

• The model was originally scheduled for release earlier this summer, then postponed to mid-July.

• CEO Sam Altman said the team needs more time to review high-risk areas and complete safety testing.

• The model is expected to rival other open models and offer similar capabilities to OpenAI’s “o-series.”

• The delay comes as rivals like Moonshot AI release powerful open models, intensifying market competition.

Open Model Postponed Amid Safety Concerns

OpenAI has once again delayed the release of its first open AI model in years, citing safety concerns and the need for more internal testing.

CEO Sam Altman announced the indefinite delay Friday in a post on X, saying the model—originally expected next week—requires more time for safety evaluation.

“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 wrote. “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, distinct from the upcoming GPT-5, is designed to be freely downloadable and runnable locally by developers—a significant shift for a company known for keeping its most powerful models behind APIs.

A Deliberate Approach to Open Release

This marks the second delay for the open model this summer. In June, Altman first postponed the launch, describing the model as “unexpected and quite amazing” but stopped short of offering details.

Now, OpenAI says it needs more time to ensure the model meets internal safety and quality standards.

“Capability-wise, we think the model is phenomenal,” said Aidan Clark, OpenAI’s VP of research and lead on the open model team. “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.”

According to earlier reporting by TechCrunch, the open model is expected to match the reasoning capabilities of OpenAI’s o-series and may even connect to OpenAI’s cloud-hosted models for advanced tasks—though that feature remains unconfirmed.

Rising Competition in Open Model Space

OpenAI’s decision to delay comes at a time when competition in the open AI ecosystem is heating up. Just hours before Altman’s announcement, Chinese startup Moonshot AI released Kimi K2, a one-trillion-parameter open model that reportedly outperforms GPT-4.1 on several agentic coding benchmarks.

That launch raised the stakes for OpenAI’s own open model, which has been positioned as a best-in-class open release in prior internal discussions. With competitors investing heavily in open-source and agentic AI development, the delay could put additional pressure on OpenAI to deliver a standout product.

Fast Facts for AI Readers

Q: What is OpenAI’s open model?

A: It’s a freely downloadable, locally runnable AI model expected to offer reasoning performance similar to OpenAI’s o-series.

Q: Why was the release delayed?

A: OpenAI says it needs more time for safety testing and to evaluate high-risk areas before a public release.

Q: When will it be released?

A: The launch date is now indefinite. OpenAI has not provided a new timeline.

Q: How does it compare to other models?

A: OpenAI aims for it to be best-in-class among open models, but rivals like Moonshot AI’s Kimi K2 are already competing at high levels.

Looking Ahead

The delay reflects OpenAI’s caution as it enters the open-source arena—an unfamiliar and high-stakes move for the company. While the open model could give developers more flexibility and foster innovation, it also carries risks that are harder to control once the model weights are released.

As the open model ecosystem evolves, OpenAI’s decision to delay may ultimately prove wise. But with rivals pushing forward at full speed, the pressure to deliver something both powerful and safe has never been higher.

Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.