At Baidu World 2026 held today, Baidu founder Robin Li officially launched a new large model service platform — “Token Exchange.” Addressing the pain points commonly faced by enterprises and developers when choosing among a huge number of large models — “difficulty choosing” and “lack of trust” — the platform innovatively adopts the operating principles of a stock exchange, aiming to provide the industry with a neutral and trustworthy venue for model trading and distribution.
An Analogy to a Stock Exchange: Aggregating High-Quality Resources and Model Providers
In day-to-day development and commercial deployment, enterprise users often find that large models on the market come in many varieties, at different prices, from uneven sources, and with actual performance that is often hard to gauge.
The positioning of “Token Exchange” is precisely intended to solve this pain point. It is like a stock exchange for the large model industry: on the one hand, it attracts and aggregates high-quality models and model providers; on the other hand, it creates a convenient and trustworthy procurement and invocation channel for enterprises and individual developers, enabling users to more confidently select and use model resources suited to their own business needs.
Strict Admission Control and Continuous Monitoring: Firmly Opposing False Parameter Claims and Model Substitution
To ensure the quality and reliability of model resources on the platform, “Token Exchange” has built a strict screening and monitoring mechanism.
When resources are first listed, the platform conducts comprehensive admission testing, covering multiple stages such as the model itself, fingerprint comparison, model capability testing, performance metric testing, and availability testing. On this basis, the platform also continuously monitors model quality parameters, using technical means to effectively prevent industry malpractices such as “model substitution” or “false parameter claims,” ensuring that every model resource has solid quality assurance.
Establishing Public Credit Records: Multidimensional Scoring and Dynamic Risk Alerts
In addition to upfront admission testing, “Token Exchange” has also established a transparent credit evaluation system.
The platform comprehensively scores model providers across multiple dimensions, including service quality, fulfillment, security compliance, and commercial performance, and awards corresponding credit rating labels. Once the credit of certain model providers declines, the platform proactively alerts users as soon as possible, and all evaluation and scoring data ultimately becomes publicly accessible credit records.
According to data disclosed at the conference, within just one month of launch, “Token Exchange” has supplied more than 100 resources, and its average daily Token invocations have exceeded 70 billion, demonstrating strong market growth momentum.