Baidu has unveiled the Ernie 5.5 Flash Preview open-weight foundation model, and launched Token Exchange, a marketplace for AI model services.
Ernie 5.5 Flash Preview has 280 billion total parameters and activates 14 billion per token, with a 1M context window. The model API is available on Baidu AI Cloud’s Qianfan platform, and the weights will be open-sourced by November 25.
API pricing is RMB1 per million input tokens and RMB4 per million output tokens, while cached input costs RMB0.04 per million tokens. The model is available for free on OpenRouter.
Token Exchange lets customers compare model services side by side with clear pricing and performance data. It checks each resource against model fingerprints and tests capability, performance and availability. Providers are rated through regular spot checks.
In just one month, the marketplace has onboarded more than 100 model resources. It handles an average of more than 50 million calls and over 700 billion tokens a day.
The launch follows Baidu’s Q2 results on August 18, when AI Cloud Infra revenue reached RMB7.3 billion, up 50 percent from a year earlier.
“If AI is the new electricity, then the intelligent organisation is the new factory of the AI era, where productivity truly happens,” said Robin Li (top), Co-founder and CEO of Baidu.
Baidu also upgraded its agent lineup with the general-purpose agent DuMate, document workspace Kooko, digital human agent Baidu Yijing and self-evolving agent Famou 3.0. Inside Baidu, daily active agents rose from 5,000 to 37,000 over the past year.
On the infrastructure side, Baidu AI Cloud launched the Tianchi 1024-card Supernode. Chip unit Kunlunxin said its M100 inference chip has shipped as planned and the flagship M300 is expected to ship in 2027.
Apollo Go now covers 28 cities worldwide. It has logged more than 350 million autonomous kilometres, more than 240 million of which were fully driverless.






Leave a Reply