Model Positioning
Qianfan-Sug is a dialogue prediction model developed in-house by Baidu's Qianfan platform, specialized in recommending the next question in multi-turn conversations. Baidu lists it as a text generation model in the Inference Service API V2. (ai.baidu.com)
Capabilities and Uses
The model identifies user intent from the existing conversation context and predicts questions the user may ask next. It can be used in scenarios such as intelligent customer service and e-commerce to suggest follow-up questions to users. (ai.baidu.com)
Limitations
The model is intended for dialogue prediction and question recommendation; its predictions are not guaranteed to match the user's actual next question. Baidu's training documentation states that a single data item supports 8192 tokens; this statement should not be treated directly as a verified API input or output limit. (ai.baidu.com)
Release Date
The release date listed in Baidu's model update record is May 15, 2025. (ai.baidu.com)
Pricing
The domestic preset service usage prices published by the Baidu AI Cloud Qianfan community are: input 0.0003 Chinese yuan (CNY) per thousand tokens, output 0.0006 Chinese yuan (CNY) per thousand tokens. These are the prices in the official announcement; refer to the Qianfan console for current actual billing. Date checked: November 18, 2026. Official pricing announcement. (qianfan.cloud.baidu.com)
Official Sources
- Baidu Qianfan model update record: Direct access to the specified page timed out during this check; the release information can be verified in Baidu's official mirror record. ()
- Baidu Qianfan model training documentation: Describes the model's positioning and training data length. (ai.baidu.com)