A clean API for the Inferfare model.

01 / SET UP

Get ready in three steps

02 / MAKE A REQUEST

Chat Completions

Send a standard OpenAI-compatible request to POST https://staging-api.inferfare.com/v1/chat/completions. Responses use the familiar choices[0].message.content shape.

curl https://staging-api.inferfare.com/v1/chat/completions \
  -H "Authorization: Bearer $INFERFARE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "inferfare",
    "messages": [
      { "role": "user", "content": "Write a launch note for Inferfare." }
    ]
  }'

03 / KEEP YOUR SDK

Use the OpenAI SDK you already have

Set the SDK base URL to https://staging-api.inferfare.com/v1 and use an Inferfare virtual key. No provider secret belongs in your application code.

04 / DISCOVER MODELS

List models available to your key

Use GET /v1/models to confirm that the key is valid and see the model IDs your application can call. The included model is inferfare. Models you add in the Inferfare workspace remain private to that workspace and are not exposed by your public virtual key.

curl https://staging-api.inferfare.com/v1/models \
  -H "Authorization: Bearer $INFERFARE_API_KEY"

05 / VIRTUAL KEYS

Use a key per app or environment

Virtual keys identify the calling application and apply your Inferfare credit. Set an environment label, expiry, model allowlist, and RPM/TPM/cap per key. Generate a separate key for production, staging, agents, or internal tools. An account can hold up to 5 active keys. Revoke a key immediately if it is no longer needed.

06 / MODEL CONNECTIONS

Add your own provider or deployment

In Settings → Models, add a BYOK connection from OpenAI, Anthropic, Google AI Studio, Groq, Mistral, or an OpenAI-compatible endpoint. Inferfare validates the connection before saving it. Your provider credential stays server-side and that provider bills usage.

HOSTED CATALOG

Hosted model IDs

When Inferfare enables host keys, these IDs draw prepaid credit: hosted-gpt-4o-mini, hosted-gpt-4o, hosted-claude-sonnet, hosted-gemini-flash, hosted-llama-70b, hosted-mistral-small. Until then they stay listed and locked. Embeddings use inferfare-embed (nomic-embed-text). Retrieval can rerank excerpts per turn.

07 / CREDIT & BILLING

Understand model credit

Inferfare sells prepaid credit only. Pay $25, $50, $100, or any whole-dollar amount from $1 to $1000. You get that same amount in Inferfare credit. There is no monthly fee and no automatic renewal. Credit does not expire. Use it on inferfare and on hosted catalog models. BYOK stays on the provider bill.

Credit is added after the payment gateway confirms the charge. Invoices appear in Settings → Credit. Inferfare emails once at 80% used unless you turn that off in Settings → Account.

CURSOR / OPENAI-COMPATIBLE CLIENTS

Connect a compatible client

In your client's custom OpenAI provider settings, use Base URL https://staging-api.inferfare.com/v1, your Inferfare virtual key, and model ID inferfare. The same values work for other OpenAI-compatible tools.

08 / TROUBLESHOOT

Common responses