API Guide
Request surfaces for adding Ando as an inference provider.
Use this guide when your app provider harness needs the exact request surfaces behind Ando. For hosted apps, call these routes from your backend with the user's Ando Virtual Key.
Base URL
https://inference.andoai.xyz/v1
Account-owned provider routes
These routes use:
Authorization: Bearer <user-virtual-key>GET /v1/modelsValidate the user's key and populate the model choices allowed for that Ando connection.
POST /v1/chat/completionsPrimary OpenAI-compatible chat route for app builders.
POST /v1/completionsLegacy text completion route for clients that still use the older shape.
POST /v1/responsesBuffered text Responses subset for clients that use the Responses-style interface.
POST /v1/thinkAndo reasoning route for flows that intentionally use the Ando-specific path.
Chat request
curl https://inference.andoai.xyz/v1/chat/completions \
-H "Authorization: Bearer $ANDO_VIRTUAL_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "Qwen/Qwen3-8B-AWQ",
"messages": [
{"role": "user", "content": "Write a short welcome message."}
],
"max_tokens": 128,
"temperature": 0.2
}'Use the model IDs returned by GET /v1/models where possible. Pinned models
are helpful in tests, but production apps should respect the user's current
Ando connection.
Accountless agents
Tempo MPP is separate from the user-owned provider path. Use it only when an agent should pay at request time without an Ando Virtual Key.
POST /v1/mpp/chat/completionsAccountless paid inference through Tempo MPP. Do not send bearer keys to this route.