The SiliconCloud platform recently announced the launch of the Batch Inference function of DeepSeek-R1& V3API, providing users with more efficient large-scale data processing solutions. Through this function, users can send batch requests to the platform, get rid of the limitations of real-time inference rate, and complete data processing tasks within 24 hours. The launch of this feature marks further innovation by SiliconCloud in the field of artificial intelligence services.
This update not only brought about functional improvements, but also accompanied by a significant price reduction. The price of DeepSeek-V3 batch inference is reduced by 50% compared to real-time inference, saving users a lot of costs. In addition, from March 11 to March 18, DeepSeek-R1 batch reasoning launched a limited-time discount, with the input price only 1 yuan/million Tokens and the output price is 4 yuan/million Tokens, a decrease of 75%. This preferential policy undoubtedly provides users with a more cost-effective choice.
The launch of the batch reasoning function is designed to meet users' needs when handling large-scale tasks such as generating reports and data cleaning. Compared with real-time inference, batch inference is more suitable for scenarios that do not require immediate response, such as data analysis, model performance evaluation, etc. Through this feature, users can complete complex data processing tasks in a shorter time while enjoying lower costs.
It is worth mentioning that DeepSeek-R1& V3API has previously supported a number of functions, including Function Calling, JSON Mode, Prefix and FIM. In addition, the TPM (token processed per minute) cap of the Pro version of DeepSeek-R1& V3API has been increased from 10,000 to 1 million, further enhancing its processing power. These updates make the SiliconCloud platform more competitive in the field of artificial intelligence services and provide users with stronger technical support.