Baidu launches "Ciyuansuo" large model trading platform
At the World 2026 conference, Baidu launched "Ciyuansuo," a large model service trading platform positioned as an exchange for the large model field, relying on the Qianfan foundation to match model resources with service providers. Li Yanhong said that the industry's "token relay stations" suffer from problems such as passing off low-spec versions as Pro versions and the inability to verify model authenticity, making enterprise model selection difficult. He also mentioned that the revenue of Baidu's Qianfan platform's open-source model-related business has achieved more than tenfold growth.
Why read this
AI score 58
Baidu has turned model trading into an exchange format, allowing readers to judge how model selection and trusted invocation steps will be platformized.
Topics
Body · AI translation
Related Topics top start
Topic: Baidu World Conference 2026 Related Topics top end Futures Promotion begin Futures Promotion end Miaopai begin Miaopai end Video Player start Video Player end Market Chart begin Market Chart end
Sina Tech News, November 12 - At the opening ceremony of the Baidu World 2026 conference held today, Baidu founder Robin Li analyzed the industry chaos surrounding third-party model aggregation "token relay stations" and released a new large-model service trading platform, the "Token Exchange," positioned as the "stock exchange" of the large-model field and dedicated to solving the industry pain points of enterprises and developers finding it difficult to choose models and lacking trust in models.
Robin Li said that in the first nine months of this year, new models were released intensively across the industry, with multiple vendors launching new models on a single day becoming the norm. Demand for large-model calls also surged in tandem, and revenue from open-source model-related business on Baidu's Qianfan platform achieved more than tenfold growth. Amid the explosive demand, computing power resources remained persistently scarce and computing power prices rose. Many model original manufacturers, constrained by computing power bottlenecks, found it difficult to fully accommodate all customers' API call demands. At the same time, intelligent computing centers across the country accelerated construction and deployment, inference efficiency continued to improve, and the industry is in a stage of rapid expansion.
But behind the industry's prosperity, new problems have emerged: as long as an enterprise has computing power, it can provide model call services externally, and a large number of "token relay stations" have consequently sprung up. Such platforms aggregate various model resources and open calls to developers, yet do not take responsibility for model performance, service stability, or interface response speed, and there are even cases of passing off lower-spec versions as Pro advanced versions, with no way to verify model authenticity.
"When developers and enterprises implement business applications, they fall into the dilemma that the more models there are, the harder it is to choose. The same model has vastly different quotes, model quality varies unevenly, and after selection they still feel uneasy and do not dare to directly put it into business use," Robin Li pointed out, saying this has become a real obstacle to the industrial implementation of large models.
Based on this industry contradiction, Baidu launched the "Token Exchange" large-model trading platform. Analogous to the logic of stock exchanges screening high-quality listed companies, the Token Exchange will gather high-quality model resources and model service providers, enabling enterprises and developers to conveniently find, purchase, and use trustworthy large-model services, and filling the shortfall in the trustworthy transaction segment of the large-model industry.
According to the introduction, the Token Exchange will rely on the Qianfan foundation, focus on solving the industry pain points of mixed model sources and uncontrollable service quality, and build a trustworthy matchmaking transaction channel for both supply and demand sides. (Wen Meng)
news_keyword_pub,stock, iframe below main text iframe below main text General statement General statement QR code at end start QR code at end start
QR code at end start QR code at end start Editor name and work code
Editor name and work code