Providers

Caravan speaks the OpenAI chat-completions dialect to every backend through one engine (CaravanProviders.Openai_compatible), configured by a single data table (CaravanProviders.Registry). Run caravan providers to see this table live, with key status for your environment.

Name	Kind	Key env var	Default model
`ollama`	local	—	`llama3.2`
`llama_cpp`	local	—	`default` (whatever is loaded)
`vllm`	local	—	`default`
`lmstudio`	local	—	`default`
`openai`	cloud	`OPENAI_API_KEY`	`gpt-4o-mini`
`anthropic`	cloud	`ANTHROPIC_API_KEY`	`claude-sonnet-5`
`groq`	cloud	`GROQ_API_KEY`	`llama-3.3-70b-versatile`
`openrouter`	cloud	`OPENROUTER_API_KEY`	`meta-llama/llama-3.3-70b-instruct`
`together`	cloud	`TOGETHER_API_KEY`	`meta-llama/Llama-3.3-70B-Instruct-Turbo`
`deepseek`	cloud	`DEEPSEEK_API_KEY`	`deepseek-chat`
`mistral`	cloud	`MISTRAL_API_KEY`	`mistral-small-latest`
`gemini`	cloud	`GEMINI_API_KEY`	`gemini-2.0-flash`
`xai`	cloud	`XAI_API_KEY`	`grok-3-mini`
`cerebras`	cloud	`CEREBRAS_API_KEY`	`llama-3.3-70b`
`github_models`	cloud	`GITHUB_TOKEN`	`openai/gpt-4o-mini`
`nvidia`	cloud	`NVIDIA_API_KEY`	`meta/llama-3.3-70b-instruct`

Model names drift faster than any table: caravan models (live, per provider) is authoritative, and the free-tier guide covers the zero-cost entries in detail.

Key resolution order

The provider's environment variable (ANTHROPIC_API_KEY, …) — preferred;
[api_keys] <provider> = "…" in ~/.caravan/config.toml (file is 0600);
legacy openai_api_key top-level key (openai only).

A model for every weight class

caravan providers --ladder prints a curated pick per size:

Class	Suggestion	Note
tiny ~1B	`ollama / llama3.2:1b`	runs on a laptop CPU
small ~4B	`ollama / qwen3:4b`	fast local reasoning
medium ~20B	`ollama / gpt-oss:20b`	strong local, ~16 GB
large ~70B	`groq / llama-3.3-70b-versatile`	open weights, hosted fast
frontier	`anthropic / claude-sonnet-5`, `openai / gpt-4o`, `gemini / gemini-2.5-pro`

Notes and caveats

Anthropic and Gemini are reached through their official OpenAI-compatible endpoints, so tool calling and streaming work through the same code path as everyone else. A few provider-specific parameters (e.g. Anthropic's fine-grained thinking controls) are not exposed through that compatibility surface; if you need them, add a bespoke provider module — the PROVIDER signature is four functions.
Any other OpenAI-compatible server (text-generation-inference, llamafile, a lab-internal gateway): use --base-url (or base_url in the config) together with -p openai, plus OPENAI_API_KEY if the gateway wants auth.
TLS: certificates are verified against the system CA store, with hostname checking. CARAVAN_TLS_INSECURE=1 disables verification for self-signed lab endpoints (a warning is printed).