Nvidiahas published an official blog post and “RTX 50 Series GPUs The fastest option for Deepseek R1” did it.
In this regard, “The Deepseek R1 family, which was recently released, created a new excitement in the artificial intelligence world. GeForce RTX 50 Series GPUs can run the Deepseek model family faster than all other solutions in the PC market with artificial intelligence power until 3,352 trillion processing per secondNvidia, who made the statement, When the low education cost of Deepseek R1 emerged, a gigantic stock market loss of $ 600 billion had lost. For a long time, the company has offered a convenient application to run LLM in systems for a long time, said that there are more than 100 million RTX graphics card PC worldwide in the last share, and you can see it right below. Radeon shared the 7900 XTX table:
You may be interested in
Microsoft had taken a step for Deepseek before that. Microsoft’s Azure AI Foundry platform currently includes GPT-4, Mistral AI and Meta-Llama 3 as well as Deepseek R1 model developed in China. This was an interesting development because OpenaI and Microsoft are currently investigating whether Deepseek is currently using the API provided by OpenAI to train their own models. Sources, which have knowledge about the issue, say that Microsoft security researchers have found that a large amount of OpenAI data, which is believed to be linked to the Chinese initiative at the end of 2024, has been leaked. In the meantime, Microsoft is planning another deepseek step.
According to the official statement 1.5 billion parameter “Deepseek-R1-Distill-Qwen-1.5BThe ”model will be available for Copilot+ PCs. Later 7 and 14 billion parameters The versions of the future model, in the first stage Snapdragon X processor can be used on Copilot+ PCs. Later as long as Intel Lunar Lake and AMD Ryzen AI 9 Processor models will also be included. The company made a statement on this issue, “These optimized models will allow developers to use the strong NPUs in Copilot+ PCs and create artificial intelligence -supported applications working efficiently on their devices.” he said.