The Deepseek R2 model supply was postponed due to the company’s dissatisfaction

The Chinese company Deepsick, despite the previous planning, is still the exact date for modeling Deepseek R2 Has not set. It seems that the CEO of the company is still not satisfied with the performance of this artificial intelligence.

According to Reuters, according to Deepsic’s CEO, Liang Vanfong, the performance of the R2 model is not satisfactory; That’s why its supply process has stopped. The R2 model, which was the successful successor to the Deepseek R1, was originally due to be released in May. The purpose of this new model was to produce better codes and the ability to reason and function in languages ​​beyond English.

Delay in the supply of the Deepseek R2 model

In the past few months, Deepsic engineers have been modifying and improving the R2 model until Liang Vanfong gives the final green light to release the model. Of course this trend has faced serious challenges; In particular, the lack of Nvidia server chips in China, which is due to US export restrictions and reduces the speed of implementation of the new model. This issue has been confirmed by The Information website quoted by the employees of large cloud service providers in China, which present dipsic models to corporate customers.

The sudden increase in demand for the R2 model can put the infrastructure of Chinese cloud service providers; Because these companies need advanced Nvidia chips to run artificial intelligence models. The new export restrictions imposed by the Trump administration in April this year have prevented the company from selling the company’s H20 chips to the Chinese market. These chips were the only artificial intelligence processors that could have been issued to China at the time.

On the other hand, Deepsic is in contact with some large cloud companies in China and offers the technical specifications of the model to help host and distribute it through the servers of these companies. Currently, most cloud customers who use the R1 model with chips H20 Nvidia They run.

Deepsic has not yet responded to Reuters’ request for comment, but it is clear that the success and supply of the R2 model depends on a number of factors, including improved model performance and market conditions.