LLM overview
I get confused the by the terminology and the ecosystem, so here are some notes.
For coding, Anthropic seems to be king. They produce two models, Claude Sonnet and Clause Opus, widely used in coding.
They also produce Claude Code, which is confusing. Claude Code is the tool that makes use of the models. You can actually swap out the models used by Claude Code. For example, you can use Claude Code with ChatGPT
Copilot is an LLM orchestrator produce by Microsoft, It’s not an LLM directly. So I guess it make more sense to compare Copilot to Claude Code, rather than any particular model.
Cursor has its own agents and models, but you can point it at Claude models. Does it make sens to point it at Claude Code? Not sure how.
LiteLLM
LiteLLM acts as a model gateway, fronted by the OpenAPI API. I think what this means is that you can plug in LiteLLM to Claude Code or Copilot and make it use any model.