OpenAI’s planned rollout of its first publicly-available AI model in years will now be delayed until later this summer, CEO Sam Altman announced on Tuesday. Originally slated for an early summer release, the model’s debut will now occur after June, Altman indicated in an update on social media.
“We are going to take a little more time with our open-weights model, which means you should expect it later this summer, though not in June,” Altman explained. “Our research team recently made an unexpected and remarkable breakthrough. We think this improvement will be very much worth the wait but requires some additional time.”
OpenAI’s open model is expected to deliver reasoning capabilities comparable to that of the company’s proprietary “o-series” AI models. Earlier, the organization outlined its ambition to surpass existing open-model offerings, including DeepSeek’s R1, which set high benchmarks in AI reasoning.
The delay comes amid increasingly fierce competition in the AI research community. Just this week, AI startup Mistral introduced its own reasoning-focused model family named Magistral, intensifying the pressure on OpenAI. Back in April, Qwen, a prominent Chinese AI research lab connected to Alibaba, also revealed a series of advanced hybrid AI models capable of detailed reasoning processes as well as quick standard responses.
Against this heightened competitive backdrop, OpenAI has also reportedly considered integrating advanced functionalities into their new open model. Among potential upgrades discussed is the ability for the model to interface directly with OpenAI’s cloud-hosted proprietary systems, allowing it to handle especially complex queries. However, it remains uncertain whether such sophisticated features will appear in the final iteration.
The release of an advanced publicly accessible AI model holds strategic significance for OpenAI, particularly when it comes to fostering goodwill and repairing relations with the academic community and developer ecosystem. OpenAI has faced criticism in past years for retaining most of its advances behind closed doors, a practice Altman himself has previously described as placing the company “on the wrong side of history.” Moving forward with a robust, feature-rich open model is likely part of a broader attempt to shift public perception, reaffirming OpenAI’s commitment to open source collaboration.
