OpenAI is preparing to launch a new open-source AI model, marking its first such release in approximately five years. This model will be freely downloadable and not restricted behind an API, with an anticipated release in early summer. The company aims for this model to outperform existing open models from competitors like Meta and DeepSeek.
A notable feature under consideration is the model’s ability to connect with OpenAI’s cloud-hosted models to enhance responses to complex queries. This handoff capability would allow the open model to access more powerful resources via the OpenAI API, providing substantial computational support when needed. The concept was proposed during a recent developer forum and has gained traction within the company.
This integration mirrors approaches like Apple’s AI capabilities, which combine on-device models with cloud-based processing. For OpenAI, this strategy could generate additional revenue and attract more developers to its premium ecosystem.
Details regarding pricing, rate limits, and the full scope of the handoff feature are still being finalized. The model is being developed from scratch and is expected to underperform OpenAI’s o3 but surpass DeepSeek’s R1 reasoning model on certain benchmarks.
