Quickstart

Three steps: create a key, export it, send a request with any OpenAI-compatible SDK.

1. Create an API key#

Sign in with Google or GitHub, then create a key from your account. Keys start with sk-hd-v1- and are shown once — store yours in a secret manager or an environment variable:

export HEIMDALL_API_KEY="sk-hd-v1-..."

2. Pick a model#

Browse the model list and copy the id of the model you want, for example deepseek/deepseek-v4-flash-0731-abliterated. Every model page shows the providers serving it, their prices and measured performance.

3. Send a request#

curl https://api-staging.impossiblecarrot.cc/api/v1/chat/completions \  -H "Authorization: Bearer $HEIMDALL_API_KEY" \  -H "Content-Type: application/json" \  -d '{    "model": "deepseek/deepseek-v4-flash-0731-abliterated",    "messages": [{ "role": "user", "content": "Hello" }]  }'

The response is a standard chat.completion object. Heimdallm adds two fields: provider — the name of the provider that served the request — and usage.cost, the exact amount charged in USD:

{
  "id": "gen-9f2c7e1a4b6d",
  "object": "chat.completion",
  "model": "deepseek/deepseek-v4-flash-0731-abliterated",
  "provider": "ImpCC",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "Hello! How can I help?" },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 12,
    "completion_tokens": 34,
    "total_tokens": 46,
    "cost": 0.0000148
  }
}

Every response also carries an X-Generation-Id header. That id is the id of the completion, and you can look up the full billing record for it later with GET /generation.

Streaming#

Set "stream": true to receive server-sent events. Chunks follow the OpenAI streaming format; the final chunk before data: [DONE] carries the usage totals, including cost.

curl -N https://api-staging.impossiblecarrot.cc/api/v1/chat/completions \
-H "Authorization: Bearer $HEIMDALL_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "model": "deepseek/deepseek-v4-flash-0731-abliterated", "stream": true,
      "messages": [{ "role": "user", "content": "Hello" }] }'

The OpenAI SDKs stream the same way they do against OpenAI:

stream = client.chat.completions.create(
    model="deepseek/deepseek-v4-flash-0731-abliterated",
    messages=[{"role": "user", "content": "Hello"}],
    stream=True,
)
for chunk in stream:
    if chunk.choices and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

See the streaming guide for the frame-by-frame anatomy, keep-alives and how mid-stream errors are delivered.

Next steps#

• How Heimdallm works — the marketplace's moving parts in one page.
• Choosing a model — reading prices and measurements to pick between endpoints.
• Models & routing — fall back across models, pin or exclude providers, route by price or speed.
• Pricing & billing — how requests are priced, credits and spend limits.
• Your account — balances, the ledger, keys and routing defaults.
• Errors & rate limits — what to expect when a request cannot be served.
• Chat playground — try any model from the browser, on an API key you hold; requests are billed to that key's workspace.
• Coding playground — build software with an AI agent on a machine of your own, billed by the minute from your credit.