Model providers¶
Each provider guide shows how to configure a model service for Snodo. Model IDs are associated with a provider in this order: Snodo's built-in model-prefix rules, the configured provider block whose name comes before the first /, then LiteLLM's own provider detection.
For credentials, Snodo checks api_key first (including encrypted @keys/<provider>.key references), then api_key_ref, then the environment variable named by api_key_env. A provider block's base_url sets its API endpoint. litellm_provider selects the LiteLLM provider format used to route its models; model_routes can select that format for matching model IDs, taking precedence over the block default. Run snodo models --check to confirm a setup.
| Guide | Use this when |
|---|---|
| Cloudflare Workers AI | You want to use models served by Cloudflare Workers AI. |
| Cloudflare AI Gateway | You want to route models from multiple vendors through an AI Gateway. |
| Ollama and OpenAI-compatible servers | You want local Ollama or another OpenAI-compatible endpoint. |
| OpenCode Go | You want to use models through the OpenCode Go gateway. |
| Anthropic | You want to connect to Anthropic's Claude models. |
| OpenAI | You want to connect to OpenAI models. |
| You want to use Google's Gemini models. | |
| OpenRouter | You want to access models through OpenRouter. |
| DeepSeek | You want to connect to DeepSeek models. |