Teepin Inference
Open models through one OpenAI-compatible API, paid per token, with confidential models that prove where they run.
Teepin Inference serves open models behind one API. You choose a model per request, pay per token from your prepaid credit, and can move to another provider without rewriting your code, because the API follows OpenAI's.
Models
List the models your key may call. Each one shows its context window, what it supports, its price per million input and output tokens, and whether it is confidential.
curl https://api.teepin.com/v1/models \
-H "Authorization: Bearer $TEEPIN_API_KEY"A model's id (for example teepin/glm-5.3) is what you send as model. It can contain a /.
What a model entry holds
| Field | Meaning |
|---|---|
id | The name to send as model |
display_name | A readable name |
context_window | Tokens the model can take in, when known |
supports_tools, supports_vision, supports_audio | What the model accepts |
pricing | input_per_million_tokens and output_per_million_tokens, in USD |
confidential | Served in a hardware enclave. See Confidential models |
attestation | Only on a confidential model that is not fully verified right now: a level and a warning to show |