AI News: Tencent’s New AI Model Challenge DeepSeek R1 in Response Speed

On Thursday, Chinese tech giant Tencent released a new AI model, “Hunyuan Turbo S.”

Tencent claims that this new can answer queries faster than the globally renowned DeepSeek’s R1.

Key Points

• The company explained that the model is capable of near-instant replies within a second by doubling the output speed and reducing the delay of the first word by almost 44%.
• The Turbo S model can respond to queries within a second by using a short thinking chain that resembles human intuition.
• The company stated that Hunyuan Turbo S demonstrated performance comparable to leading models on the market, such as DeepSeek-V3, OpenAI’s GPT-4o, and Anthropic PBC’s Claude, in benchmarks for math, reasoning, and knowledge.
• To create the model, Tencent used a Hybrid-Mamba-Transformer fusion to reduce the computational complexity and KV-Cache of the model’s Transformer architecture.
• Tencent also mentioned that the usage costs of the new Turbo S were significantly lower than its previous iterations, highlighting how DeepSeek’s open-source low-pricing strategy has compelled other leading Chinese AI firms to charge users less.
• Tencent added that as a flagship model, Turbo S will become central for foundation models in the future for inference, text, and code generation.

Background

Tencent, a leading Chinese tech giant, has been a key player in AI development, with models focusing on natural language processing, gaming, and cloud computing.

Tencent, built its empire through WeChat, online gaming, and strategic investments across global tech companies.

Before its latest AI model, Hunyuan Turbo S, Tencent introduced Hunyuan, a large language model with strong reasoning, writing, and coding capabilities.

Hunyuan emphasized efficiency and multimodal AI applications, competing with global counterparts.

Tencent continues advancing AI to enhance business, entertainment, and cloud-based solutions

News Gist

On Thursday, Tencent launched Hunyuan Turbo S, an AI model with near-instant replies, outperforming DeepSeek’s R1.

It enhances speed using Hybrid-Mamba-Transformer fusion, reduces costs, and rivals GPT-4o and Claude in benchmarks.

Turbo S will be central to Tencent’s AI strategy.