OpenAI Delays Open Model Release - Again

OpenAI Delays Release of Open Model for Further Safety Checks

Sam Altman, CEO of OpenAI, announced on Friday that the release of the company’s open model has been postponed. This follows a previous delay announced earlier this summer.

The initial plan was to launch the model next week, but Altman stated that additional safety testing and review of potential risks are necessary.

Altman's Statement on X

“We need time to run additional safety tests and review high-risk areas,” Altman explained in a post on X. “We are not yet sure how long it will take us.”

He further emphasized the irreversible nature of releasing the model weights, stating, “Once weights are out, they can’t be pulled back. This is new for us and we want to get it right.”

Significance of the Open Model Release

OpenAI’s open model release was anticipated as a major event in the AI landscape this summer, alongside the expected launch of GPT-5.

Unlike GPT-5, this open model is designed to be freely downloadable and runnable locally by developers.

Through these releases, OpenAI aims to solidify its position as a leading AI lab in Silicon Valley, facing increasing competition from companies like xAI, Google DeepMind, and Anthropic.

Developer Impact and Model Capabilities

This delay means developers will experience a longer wait to access OpenAI’s first open model in years.

Reports suggest the model is expected to exhibit reasoning capabilities comparable to OpenAI’s o-series models, and to be a top performer among other available open models.

Increased Competition in the Open AI Model Space

The open AI model ecosystem has become more competitive recently.

On Friday, Moonshot AI, a Chinese AI startup, launched Kimi K2, a one-trillion-parameter open AI model that surpasses OpenAI’s GPT-4.1 in several agentic-coding benchmarks.

Previous Announcements and Internal Discussions

When the initial delays were announced in June, Altman mentioned the company had achieved something “unexpected and quite amazing,” but did not provide further details.

OpenAI's Standards for Open Source Models

Aidan Clark, OpenAI’s VP of research and leader of the open model team, stated on X Friday, “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.”

Potential Future Features

Previous reports indicated OpenAI leaders have considered enabling the open AI model to connect to the company’s cloud-hosted AI models for handling complex queries.

However, it remains uncertain whether these features will be included in the final version of the open model.