Provider catalog

The installed runtime is the canonical catalog:

odinn config provider catalog

A preset provides an authentication mode, endpoint shape, transport, environment-variable name, and optional model defaults. It does not guarantee that a provider account, endpoint, or named model is currently available.

Browser OAuth and OAuth-capable paths

OpenAI / ChatGPT
OpenRouter
Chutes

OAuth behavior still depends on the external authorization service and account.

Device or CLI authentication

GitHub Copilot
xAI device OAuth
Antigravity
Google Antigravity

These paths depend on the corresponding installed or remote authentication flow.

API-key presets

Arcee
Cerebras
Cohere
DeepInfra
DeepSeek
Featherless
Fireworks
GMI
Groq
Hugging Face
Kilo Code
LongCat
Mistral
Moonshot and Moonshot China
NVIDIA
Novita
Qianfan
Together
Venice
Volcengine and Volcengine Coding Plan
Xiaomi
xAI
Z.ai global, China, and coding variants

Provider service names, model names, URLs, quotas, and billing can change. Use the installed catalog and the provider’s current documentation rather than copying an old model name into automation.

Local and self-hosted compatibility

Ollama
LM Studio
LiteLLM
vLLM
SGLang

Start the server first and configure the exact model it exposes.

Generic compatible endpoint

Add a custom OpenAI-compatible service:

odinn config provider add <name> \
  --base-url <url> \
  --model <model> \
  --api-key-env <ENVIRONMENT_VARIABLE>

Then verify through the real runtime:

odinn onboard --verify --non-interactive

See Configure providers for authentication and secret handling.