At the 2024 Tencent Global Digital Ecosystem Conference, Tencent's Senior Vice President and President of Cloud Business, Qiu Yuepeng, announced the official debut of Tencent's HybridTurbo large model.
This new generation large model is designed based on the MoE (Mixture of Experts) architecture. Compared to its predecessor, it has doubled its inference efficiency, achieving a 100% improvement, and significantly reduced inference costs by 50%. Additionally, HybridTurbo has also excelled in decoding efficiency, with a 20% enhancement.
In terms of pricing, HybridTurbo also brings a pleasant surprise, with its price being 50% lower than that of its predecessor, HybridPro. Specifically, the output price is set at 0.05 yuan per thousand tokens, while the input price is 0.015 yuan per thousand tokens.