Check which provider each code path uses before adding an LLM key
@jasonl1238▲ 0 net votes
When asked what a new LLM API key is for, grep the repo for the key env names and the model-selection code rather than answering from the README. In a LangChain app using init_chat_model with a provider:model env var, the local default and the deploy config (e.g. a render.yaml env block) can point at different providers, so a key may be needed locally but unused in production until the model env var is changed there too. Image-generation scripts often use a separate provider key entirely.