OpenAI delays open-weight model release: summer launch to rival DeepSeek’s R1

OpenAI has postponed the launch of its much-anticipated open-weight language model, pushing the release from early summer to later this summer. CEO Sam Altman announced the delay on X, emphasizing the need for additional development time.

The open-weight model is designed to match the reasoning capabilities of OpenAI's existing o-series models, such as the recently rolled out o3-pro, which has been praised for improved performance in science, education, programming, data analysis, and writing.

"We are going to take a little more time with our open-weights model, i.e. expect it later this summer but not June," Altman stated.

Open-weight models differ from fully open-source AI by releasing only the trained parameters, allowing researchers and developers to run the models locally without cloud API dependency, while maintaining some usage restrictions.

The upcoming model aims to outperform competitors like DeepSeek's R1, a leading open reasoning model, signaling OpenAI's commitment to advancing accessible AI technology.

Despite the delay, OpenAI continues to improve its offerings; the o3-pro model, while lacking image generation and Canvas support and having slower response times than its predecessor, has been consistently preferred by expert reviewers.

As AI models grow more powerful, concerns about energy consumption rise. A typical ChatGPT query uses about 0.34 watt-hours of energy and 0.000085 gallons of water, with predictions that AI's total power use could surpass Bitcoin mining by 2025.

"In expert evaluations, reviewers consistently prefer OpenAI o3-pro over o3, highlighting its improved performance in key domains science, education, programming, data analysis, and writing," OpenAI noted.

The delay underscores the complexity of developing open-weight models that balance accessibility, performance, and responsible deployment.

OpenAI has delayed the release of its highly anticipated open-weight language model, initially expected in early summer, to later this summer. CEO Sam Altman said the model aims to rival DeepSeek’s R1 with advanced reasoning capabilities, following the rollout of the improved o3-pro model.