A model switching method and related device

By adaptively selecting and switching AI models through the model management system, the problem of large models being unable to meet the needs of complex business applications has been solved, achieving more efficient task solving capabilities and cost optimization.

CN122241034APending Publication Date: 2026-06-19HUAWEI CLOUD COMPUTING TECHNOLOGIES CO LTD
0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
HUAWEI CLOUD COMPUTING TECHNOLOGIES CO LTD
Filing Date
2024-12-17
Publication Date
2026-06-19

AI Technical Summary

⚠Technical Problem

In existing technologies, large-scale model applications struggle to meet business needs when faced with complex operations, and the fixed use of the same model leads to reduced end-to-end accuracy, increased costs, and longer response times.

⚗Method used

This paper provides a model switching method that adaptively selects and switches AI models through a model management system. It selects a suitable model based on feature matching of query requests and optimizes model parameters by combining integer programming and mixed integer programming to achieve adaptive model switching and parameter configuration.

🎯Benefits of technology

It improved the ability to solve tasks ranging from simple to complex, increased end-to-end accuracy, shortened response latency, reduced costs, and met business needs.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

A model switching method includes: receiving a first query request; extracting a first feature based on the first query request; selecting a first model from multiple AI models that matches the first feature based on the first feature, wherein each AI model has one or more matching features; using the first model to infer the first query request to obtain a first model-generated result; and providing the first model-generated result on a user interface; receiving a second query request and extracting a second feature based on the second query request; selecting a second model from multiple AI models that matches the second feature based on the second feature; using the second model to infer the second query request to obtain a second model-generated result; and providing the second model-generated result on a user interface. This achieves adaptive model switching, avoiding the fixed use of the same model, fully leveraging the advantages of each AI model, improving the ability to solve tasks from simple to complex, and meeting business needs.
Need to check novelty before this filing date? Find Prior Art