DeepSeek’s efficiency-focused V4-Flash model reached general availability on July 31, priced at 9.5 times less than its Pro sibling and outperforming it on at least one coding benchmark.
Tag: DeepSeek
Posted inTech
DeepSeek is building its own AI inference chip to cut reliance on Nvidia
China’s fastest-growing AI startup has quietly assembled a semiconductor engineering team and begun talks with design and manufacturing partners to develop a proprietary chip optimised for inference workloads.