Baidu's Wenxin X1 Large Model Officially Launches on Qianfan Platform

On April 2, Baidu's Wenxin X1 large model officially launched on the Qianfan platform, allowing enterprise users and developers to call the API upon login.

Wenxin X1 features enhanced understanding, planning, reflection, and evolution capabilities, supporting multimodal inputs. It is a comprehensive deep-thinking model with accuracy, creativity, and literary quality, excelling in Chinese knowledge Q&A, literary creation, document writing, daily conversation, logical reasoning, complex calculations, and tool invocation.

According to authoritative testing, Wenxin X1 demonstrates excellent performance in mathematics, coding, and knowledge reasoning across multiple public datasets, surpassing the upgraded DeepSeek-V3-0324. In mathematics, GSM8K dataset testing showed Wenxin X1 scoring 95.6 versus DeepSeek-V3-0324's 93.6; in code generation, MBPP dataset testing showed Wenxin X1 scoring 83.3 versus DeepSeek-V3-0324's 81.2; in knowledge reasoning, C-Eval dataset testing showed Wenxin X1 scoring 88.6 versus DeepSeek-V3-0324's 85.1; in mathematical reasoning, Aime2024 dataset testing showed Wenxin X1 scoring 78.6 versus DeepSeek-V3-0324's 57.1.

For model services, users can directly call Wenxin X1 API through Qianfan ModelBuilder, with input prices as low as 0.002 yuan per thousand tokens and output prices as low as 0.008 yuan per thousand tokens. Batch inference scenarios are also supported to meet diverse business needs.

For model development, Qianfan ModelBuilder's model distillation supports Wenxin X1 as a teacher model, enabling one-click knowledge distillation from 'thinking models' to 'lightweight models'.

For application development, Qianfan AppBuilder supports users in quickly building RAG, Agent, AI search, and workflow applications based on Wenxin X1, achieving more thorough thinking planning, more accurate tool invocation, and more comprehensive generation effects, helping enterprise-level large model applications land across various industries.

Baidu Intelligent Cloud's Qianfan large model platform is committed to providing users with full-process, one-stop AI services, with an open, easy-to-use, and low-cost platform philosophy. Enterprise users and developers can more efficiently explore large model applications, improve innovation efficiency, and accelerate the transformation of various AI applications from concept to implementation, injecting powerful momentum into the expansion and application of AI technology in more fields.