The Chinese have introduced a fast-thinking AI Turbo S – it responds faster than DeepSeek

Tencent Holdings has unveiled an AI model called Hunyuan Turbo S, which it says is faster than the DeepSeek R1, which has the ability to think. Tencent also noted that the cost of using Hunyuan Turbo S is many times lower than previous versions of its AI models, Bloomberg reports.

Image source: Tencent

Hunyuan Turbo S is capable of responding to queries within a second, which distinguishes it “from DeepSeek R1, Hunyuan T1 and other slow-thinking models that need to ‘think’ for a while before responding,” Tencent said. The company added that when tested in areas such as knowledge, mathematics and reasoning, Turbo S’s capabilities match those of DeepSeek V3, the AI ​​model that powers the DeepSeek chatbot that has surpassed OpenAI’s ChatGPT in app store downloads. To back up its claims, Tencent cited data showing that Hunyuan Turbo S can compete with DeepSeek V3 in popular AI tests.

The success of the DeepSeek R1 and V3 models, which have become widely known and attracted interest in Silicon Valley, has forced Chinese tech giants such as Tencent and Alibaba to accelerate the release of new versions of AI models.

Just days after the release of the DeepSeek R1, e-commerce giant Alibaba unveiled the Qwen 2.5-Max, which it claims outperforms the DeepSeek V3 in every way, as well as outperforms other leading models such as GPT-4o and Sonnet’s Claude-3.5.

Tencent, which along with Alibaba is one of the largest cloud infrastructure providers in China, integrated DeepSeek’s technology into its WeChat messaging app, sending its shares higher. DeepSeek’s chatbot is also widely used by government agencies across China, including in Hong Kong.

To cope with sudden changes in demand, DeepSeek this week lowered the price of its API access during nighttime hours. DeepSeek’s strategy of offering open-source solutions at low prices has forced other leading Chinese AI companies to charge users less, Bloomberg notes.