Crush provides flexible model switching solutions for different model requirements in multilingual projects:
- Dynamic switching commands: Use
crush model switch 模型名Instant LLM replacement - Project-level configuration: in
~/.crush/config.jsonDefining Language and Model Mapping Relationships in - Catwalk Support: Quickly add new models through a community-maintained model library
- hybrid deployment: Configure both cloud-based models (e.g. GPT-4) and local models (Ollama)
For example, Java projects can use Claude to handle business logic, and Python data analysis is switched to GPT-4 through the"providers"Configure blocks to automate switching.
This answer comes from the articleCrush: endpoint AI programming assistant with integrated LSP and multi-model switchingThe
May not be reproduced without permission:AI productivity tools " Enabling Intelligent Switching of Models in Multilingual Project Development
English 
简体中文 
日本語 
Deutsch 
Português do Brasil