How can the big model Liu Xiaoqiang reshape his strategy under the impact of DeepSeek?

Over the past month, the global mockup market has been in turmoil due to the launch of DeepSeek, with major companies being asked about their views and strategies for this emerging force. In the domestic market, six large-scale model startups that have become unicorns - Zero One World, Baichuan Intelligence, Stepping Stars, Wisdom Chapter, The Dark Side of the Moon and MiniMax, are facing severe challenges, but their respective Coping strategies show different directions.

First of all, Kai-Fu Lee, the founder of Zero One World, said earlier that they no longer pursue the training of super-large models, but focus on the development of lightweight models. On February 14, Zero-1000 Things and Suzhou High-tech Zone jointly established an industrial big model base, focusing on manufacturing, finance, medical care and other fields, and committed to pushing big model technology from laboratories to practical applications.

Within just a few days after the release of DeepSeek, Baichuan Intelligent launched the full-scene reasoning model Baichuan-M1-preview, which has the capabilities of language, vision and search, and focuses on transforming into the medical industry and launches "AI" The Pediatrician project shows its rapid implementation in practical applications.

Jieyuexingchen chose to launch two new models on the day of DeepSeek's release to enhance its market competitiveness. In addition, its application "Yuewen" has also launched new functions, aiming to help users develop applications without code and further improve user experience.

Tang Jie, the founder of Zhipu Huazhang, emphasized at the international summit that the large language model of the future will become the core of daily life and work, and will collaborate with humans to complete complex tasks through independent learning and collaborate with humans to show its foresight for the future.

The Dark Side of the Moon was launched shortly after the release of DeepSeek, the Kimi k1.5 multimodal thinking model was launched to improve the model's context understanding ability and strategy optimization, indicating its continued efforts in technological innovation.

Although the six startups did not respond directly to DeepSeek, their practical actions and product innovations are already actively responding to this impact and demonstrating their ability to adapt to market changes.