AB
AiBoss
project

Wenxin Big Model X1 - Baidu's Deep Thinking Model

Wenxin Big Model X1 is a deep thinking model launched by Baidu. It possesses "long thought chains" and excels in Chinese knowledge-based question answering, literary creation, and logical reasoning. X1 adds multimodal capabilities, enabling it to understand and generate images, and call upon tools to generate...

What is Wenxin Big Model X1?

Wenxin X1 is a deep thinking model launched by Baidu. It possesses a "long thought chain" and excels in Chinese knowledge-based question answering, literary creation, and logical reasoning. X1 adds multimodal capabilities, enabling it to understand and generate images, and call upon tools to generate rich content such as code and charts. Based on key technologies such as progressive reinforcement learning and end-to-end training of thought chains and action chains, and through joint optimization with PaddlePaddle, the reasoning cost has been significantly reduced.

Main functions of Wenxin Large Model X1

  • Deep thinking abilityAs a deep thinking model, Wenxin Big Model X1 possesses a "long thinking chain" and excels in Chinese knowledge Q&A, literary creation, logical reasoning, and other areas.
  • Multimodal capabilitiesIt adds multimodal capabilities, enabling it to understand and generate images.
  • Multi-tool calling capabilityIt can call upon tools to generate rich content such as code and charts. It already supports multiple tools including advanced search, document Q&A, image understanding, AI drawing, code interpreter, web link reading, TreeMind tree diagrams, Baidu Scholar search, business information query, and franchise information query.

Technical Principles of Wenxin Large Model X1

  • Training and optimization techniquesThe Wenxin Big Model X1 employs key technologies such as progressive reinforcement learning, end-to-end training based on thought chains and action chains, and a multi-dimensional unified reward system. Through joint optimization between PaddlePaddle and Wenxin, it achieves extreme optimization across the entire chain from compression, inference, and service deployment, significantly reducing inference costs.
  • Knowledge EnhancementThe Wenxin Big Model X1 continues the knowledge enhancement technology of the Wenxin series. By integrating large-scale knowledge graphs and massive amounts of unstructured data, the model can break through the bottleneck of unified expression of heterogeneous data and achieve deep understanding and generation of knowledge.
  • Pre-training and fine-tuningThe Wenxin X1 model employs a pre-training plus fine-tuning training paradigm. In the pre-training phase, the model learns language knowledge and rules from massive amounts of unlabeled data through self-supervised learning; in the fine-tuning phase, the model is optimized based on a small amount of labeled data to adapt to the needs of specific tasks.

How to use the Wenxin large model X1

  • Use via Wenxin Yiyan official websiteThe Wenxin Large Model X1 is now available on the Wenxin Yiyan official website. Individual users can visit the website directly to experience it for free.
  • Select ModelSwitch to the Wenxin X1 model and then use it to have a conversation with Wenxin Yiyan.
  • Select the calling toolUse tools such as code interpreters, document readers, image comprehension, and image generation.

Price of Wenxin Large Model X1

  • The API call price of Wenxin Big Model X1 is about half that of DeepSeek-R1, with an input price of 0.002 yuan/thousand tokens and an output price of 0.008 yuan/thousand tokens.

Application scenarios of Wenxin Large Model X1

  • Literary creationIt helps writers and creators quickly conceive plots, generate story frameworks, or continue texts, thus inspiring creative ideas.
  • Copywriting generationWe provide high-quality copywriting for the advertising, marketing, and public relations industries, generating attractive headlines, slogans, or product descriptions based on user needs.
  • Academic SearchIt assists researchers and students in conducting academic research, quickly locates relevant literature, provides research ideas and analytical methods, and improves the efficiency of academic research.
  • Code generationIt helps developers generate code snippets, debug code, or optimize code structure, thereby improving programming efficiency and reducing development costs.
  • Data AnalysisCombine data visualization tools to quickly analyze data trends and generate reports to support decision-making.