The Release of OpenAI’s Open Model has Been Postponed

OpenAI’s first open model in years is now expected to launch later this summer, rather than in June, CEO Sam Altman shared in a post on X on Tuesday.

“We’re taking a bit more time with our open-weights model, so you can expect it later this summer—not in June,” Altman wrote. “Our research team made an unexpected and remarkable breakthrough, and while it needs a little more time, we believe it will be well worth the wait.”

OpenAI’s Open Model Aims to Outperform Competitors, But Faces Delay

OpenAI had initially planned to release its open model in early summer. The model is expected to feature reasoning abilities comparable to those in the o-series. OpenAI’s goal is for this open model to outperform other open-source reasoning models, including DeepSeek’s R1.

Since OpenAI announced plans to launch an open model, competition in the field has intensified. On Tuesday, Mistral—a lab known for releasing open models—unveiled its first AI reasoning model family, Magistral. Back in April, China-based AI lab Qwen introduced a set of hybrid reasoning models capable of switching between deep reasoning and faster, more straightforward responses.

OpenAI Explores Advanced Features to Strengthen Open Model’s Capabilities

To stay competitive, OpenAI has not only focused on boosting performance in benchmarks but has also explored adding advanced features to its open model. According to TechCrunch, company leaders have considered allowing the model to connect with OpenAI’s cloud-based systems for handling more complex tasks. However, it’s still uncertain whether these capabilities will be included in the final version.

The launch of OpenAI’s open model appears to be a key move in strengthening its ties with researchers and developers. Altman has acknowledged that the company has been on the “wrong side of history” regarding open-sourcing its models. To change that perception, OpenAI is under significant pressure to deliver an open model that can rival the top open-source models in the field.

Read the original article on: TechCrunch

Read more: No AI-Powered, Personalized Siri Revealed At WWDC 2025