2025年1月30日 · The DeepSeek-R1 NIM microservice can deliver up to 3,872 tokens per second on a single NVIDIA HGX H200 system. Developers can test and experiment with the application programming interface (API), which is expected to be available soon as a downloadable NIM microservice, part of the NVIDIA AI Enterprise software platform.

2025年1月27日 · Chinese startup DeepSeek has debuted an AI app that challenges OpenAI's ChatGPT and other U.S. rivals, sending a shock through Wall Street.

State-of-the-art, high-efficiency LLM excelling in reasoning, math, and coding.

3 天之前 · Nvidia — the dominant player in AI chip design and, as of this morning, the world’s third-largest company by market cap — saw its stock price tumble after DeepSeek’s latest model ...

4 天之前 · Nvidia counters AMD DeepSeek AI benchmarks, claims RTX 4090 is nearly 50% faster than 7900 XTX. Latest. G.Skill intros 96GB DDR5-6800 CL32 and 32GB DDR5-6400 CL28 memory kits for Intel-based machines.

2025年1月31日 · Reportedly, DeepSeek trained its V3 model on Nvidia’s tailor-made H800 chips for China, which the U.S. banned in October 2023. This prompted Nvidia to design the H20 chips.

6 天之前 · DeepSeek’s success has come despite U.S. export controls on cutting-edge chips to China. Nvidia said in a statement to Investopedia that it "insists" customers comply with laws.

2025年1月31日 · DeepSeek was trained on 2,048 Nvidia H800 GPUs. Each node in the cluster DeepSeek trained on houses 8 GPUs connected by NVLink and NVSwitch for intra-node communication, while InfiniBand ...