Tencent Cloud Launches New Pricing Strategy for Large Models, Hybrid Series Significantly Reduced in Price and Partially Free

2024-05-23

Tencent Cloud announced a price adjustment for its highly anticipated Hybrid Model series, aiming to further reduce the threshold for enterprises and individual developers to use artificial intelligence technology. It is understood that Tencent Cloud's adjustment this time mainly focuses on the API prices of the Hybrid Model series. Among them, the Hybrid-lite model, as one of the main models, will significantly increase its API input and output total length from the current 4k to 256k, and the price will be adjusted from the original 0.008 yuan/thousand tokens to free, providing users with a broader space for use. At the same time, the API prices of the Hybrid-standard model have also been greatly reduced. Specifically, the input price has been reduced from 0.01 yuan/thousand tokens to 0.0045 yuan/thousand tokens, a decrease of 55%; the output price has been reduced from 0.01 yuan/thousand tokens to 0.005 yuan/thousand tokens, a decrease of 50%. In addition, the newly launched Hybrid-standard-256k model, with its ability to process ultra-long texts of over 380,000 characters, has its API input price reduced to 0.015 yuan/thousand tokens, a decrease of 87.5%, and the output price reduced to 0.06 yuan/thousand tokens, a decrease of 50%. In the high-end market, Tencent Cloud has also made great efforts in price optimization. The highest configuration trillion-parameter model, Hybrid-pro, has its API input price reduced from 0.1 yuan/thousand tokens to 0.03 yuan/thousand tokens, a decrease of 70%. This adjustment will greatly promote the application of artificial intelligence technology for high-end users. In addition to price adjustments, Tencent Cloud has also comprehensively upgraded its one-stop intelligent body creation and distribution platform, Tencent Yuanqi, with model resource support. Users can now distribute exclusive AI intelligent bodies to Tencent's global channels with just one click, and enjoy free Hybrid Model resources and traffic support. At the same time, users can also distribute and deploy intelligent bodies in various scenarios in API form, and the free token quota has been increased from the previous 1 million to 100 million, providing developers with richer resources and support. As Tencent's self-developed trillion-parameter model, the Hybrid Model is the first in China to adopt a Hybrid Expert Model (MoE) structure, with an overall performance improvement of 50% compared to the previous generation. According to Sullivan's evaluation results, Tencent's Hybrid Model is in the top tier of large models in China, with performance surpassing the international average. Currently, the Hybrid-pro, Hybrid-standard, and Hybrid-lite models of different versions and sizes have been opened to enterprises and individual developers in the form of APIs.