OpenAI Delays Release of Its Open-Weights Model to Late Summer

OpenAI’s first open-weight AI model in years will not be released until later this summer, according to CEO Sam Altman. In a post on X Tuesday, Altman said the delay was due to new developments within the company’s research team, describing the results as “unexpected and quite amazing.”

“[W]e are going to take a little more time with our open-weights model, i.e. expect it later this summer but not [J]une,” Altman wrote. “[O]ur research team did something unexpected and quite amazing and we think it will be very very worth the wait, but needs a bit longer.”

The model was initially slated for release in early summer and is expected to have capabilities similar to OpenAI’s reasoning-focused O-series models. The goal, according to OpenAI, is to outperform existing open reasoning models such as DeepSeek’s R1.

Competition Heats Up in the Open Model Space

The delay comes as other AI labs continue to move aggressively in the open model space. On the same day as Altman’s announcement, Mistral released a new family of open reasoning models called Magistral. In April, Chinese AI developer Qwen launched its own hybrid reasoning models, designed to toggle between slower, thoughtful outputs and faster, conventional responses.

This uptick in open model development has raised the stakes for OpenAI, which is under pressure to deliver something that’s not only open—but also competitive with top-tier alternatives.

Possible Features and Strategic Importance

Beyond strong benchmark performance, OpenAI has been exploring additional capabilities for its open model. According to a previous TechCrunch report, company leaders have discussed allowing the open model to connect with OpenAI’s more powerful, cloud-hosted tools for handling complex queries. It’s unclear whether that feature—or others still in testing—will be included in the final release.

While OpenAI has not provided a firm launch date, Altman’s comments suggest the model will arrive sometime between July and September.

The release carries symbolic weight for OpenAI’s standing in the research community. Altman has publicly acknowledged that the company has landed on the “wrong side of history” when it comes to open-sourcing AI models. Launching a strong open model is seen as a chance to correct that narrative and re-engage with developers who prioritize transparency, reproducibility, and open experimentation.

What This Means

Delaying the open model adds pressure—but also raises expectations. By promising not just openness, but innovation, OpenAI is signaling that it doesn’t just want to participate in the open ecosystem—it wants to lead it.

This release comes at a pivotal moment. Open models are becoming central to research, transparency, and global AI development—especially in countries or sectors that can’t rely on commercial APIs or proprietary systems. For many developers, researchers, and institutions, open models are more than a convenience; they’re a necessity.

With competitors like Mistral and Qwen rapidly expanding their offerings, OpenAI’s long-awaited contribution could reshape the balance between closed and open AI development. But to do so, it will need to deliver a model that’s not only technically strong, but also meaningfully accessible.

This isn’t just a test of performance—it’s a test of trust.

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.