Add your own AI models and providers

Enabling a provider

Each provider has a switch and an API key field. Turn on the ones you have keys for: OpenAI, Anthropic, Google AI Studio, Groq, DeepSeek, xAI or OpenRouter, and paste the key.

Adding models

Grape does not ship a fixed list of every provider's models, which would fall behind each time one is released. Click Add Model under a list instead, and Grape asks each provider you switched on which models it offers you, with your own key, newest first. Search, pick one, and it joins the list.

A chat model joins switched on, ready for the chat model picker. The other lists run on one model at a time, so a model you add there joins switched off while another one is on, and nothing changes until you switch it on.

Only the providers that can do a job are asked for it: transcription comes from OpenAI and Groq, images from OpenAI, Google AI Studio and xAI, and indexing from OpenAI and Google AI Studio. A custom endpoint can fill every list but transcription, and on desktop Ollama and LM Studio fill the chat and indexing lists. If Add Model says a provider did not accept the key, check the key under Providers and try again.

Picking models

Every model has its own switch, so the pickers only show what you actually use. When more than one provider is enabled, the list shows which provider each model belongs to. The trash button beside a model takes it out of the list again.

Adding a model by its id

A model a provider does not list, like one you were given early access to, can still be added. Type its id in the Add Model search exactly as the provider writes it, and choose to add it from that provider.