OpenAI’s open model is delayed

OpenAI has announced a delay in the release of its highly anticipated open model, with CEO Sam Altman revealing on X that it will now debut later this summer, after June. In his message, Altman stated, "We are going to take a little more time with our open-weights model, so expect it later this summer but not in June. Our research team achieved something unexpected and remarkable, and we believe it will be well worth the wait, though it requires additional time." Initially, OpenAI aimed for an early summer launch for this open model, which is expected to offer reasoning capabilities comparable to its existing O-series models. The company has set its sights on ensuring that this new model outperforms other open reasoning models, such as DeepSeek's R1. The competitive landscape has intensified since OpenAI first announced its plans, with Mistral, another AI lab, recently launching its own set of AI reasoning models called Magistral. In addition, this past April saw the Chinese AI lab Qwen introduce hybrid AI reasoning models capable of alternating between in-depth problem-solving and providing quick responses. Beyond enhancing performance on established benchmarks, OpenAI is exploring the inclusion of advanced features to bolster its model's competitiveness. Reports suggest that discussions have taken place regarding the potential for the open model to interface with the company's cloud-based AI models for handling complex queries, although it remains uncertain if these features will be incorporated in the final product. The development and release of OpenAI's open model are crucial for maintaining its relationship with the broader research and developer communities. Altman has previously acknowledged that OpenAI may have taken a misstep regarding the open-sourcing of its models. To amend this perception, the company faces significant pressure to deliver an open model that stands toe-to-toe with the industry's leading offerings.