Balance and request cost

Two read-only endpoints help you track spend: GET /v1/credits reads the balance of the key’s workspace, and GET /v1/generation reads what a specific request cost. Neither one is billed.

Read the balance#Link to the Read the balance section

GET https://api.realmoon.ai/v1/credits

Returns the balance of the API wallet of the workspace the key belongs to.

• Authentication: an sk-moon-… key with the inference scope (see Authentication).
• Parameters: none.
• Response: a credits object.

Example#Link to the Example section

curl https://api.realmoon.ai/v1/credits \
  -H "Authorization: Bearer $MOONAI_API_KEY"

import os
import httpx
from openai import OpenAI

client = OpenAI(base_url="https://api.realmoon.ai/v1", api_key=os.environ["MOONAI_API_KEY"])

credits = client.get("/credits", cast_to=httpx.Response).json()
print(credits["balance"])

import OpenAI from 'openai';

const client = new OpenAI({ baseURL: 'https://api.realmoon.ai/v1', apiKey: process.env.MOONAI_API_KEY });

const credits = await client.get('/credits');
console.log(credits.balance);

Response#Link to the Response section

{ "object": "credits", "balance": 1250000, "currency": "credits" }

The balance goes back up when in-flight requests are settled at their actual cost (see Hold, then actual cost). A negative balance blocks new requests until the next top-up (402 insufficient_credits).

Errors#Link to the Errors section

Only the errors common to the whole API apply: 401 invalid_api_key, 403 ip_not_allowed, 403 missing_scope, 429 rate_limit_exceeded, 503 service_unavailable. See Errors.

Read the cost of a request#Link to the Read the cost of a request section

GET https://api.realmoon.ai/v1/generation?id=gen-…

Returns the final cost and the tokens counted for a request made in the key’s workspace.

• Authentication: an sk-moon-… key with the inference scope, from the same workspace as the request.
• Parameter: id, in the query string.
• Response: a generation object.

Parameter#Link to the Parameter section

Where to find the ID:

Example#Link to the Example section

curl "https://api.realmoon.ai/v1/generation?id=gen-0123456789abcdef0123456789abcdef" \
  -H "Authorization: Bearer $MOONAI_API_KEY"

import os
import httpx
from openai import OpenAI

client = OpenAI(base_url="https://api.realmoon.ai/v1", api_key=os.environ["MOONAI_API_KEY"])

raw = client.chat.completions.with_raw_response.create(
    model="anthropic/claude-haiku-4.5",
    messages=[{"role": "user", "content": "Hello"}],
)
request_id = raw.headers.get("x-moon-request-id")

cost = client.get(
    "/generation",
    cast_to=httpx.Response,
    options={"params": {"id": request_id}},
).json()
print(cost["credits_charged"])

import OpenAI from 'openai';

const client = new OpenAI({ baseURL: 'https://api.realmoon.ai/v1', apiKey: process.env.MOONAI_API_KEY });

const { response } = await client.chat.completions
  .create({ model: 'anthropic/claude-haiku-4.5', messages: [{ role: 'user', content: 'Hello' }] })
  .withResponse();
const requestId = response.headers.get('x-moon-request-id');

const cost = await client.get('/generation', { query: { id: requestId } });
console.log(cost.credits_charged);

Response#Link to the Response section

{
  "object": "generation",
  "id": "gen-0123456789abcdef0123456789abcdef",
  "model": "anthropic/claude-haiku-4.5",
  "tokens_prompt": 12,
  "tokens_completion": 41,
  "credits_charged": 0,
  "created_at": 1793836800
}

The credits_charged value is made up.

When to read the cost again#Link to the When to read the cost again section

• A request settled right away (complete response with its token count, finished media) has its final cost as soon as it responds.
• A response that was interrupted, or that had no token count, can still be settled at the actual cost reported by the provider for about twenty minutes: read /v1/generation again later.
• A video is settled when its job ends: its cost is also available in usage.credits of the video object (see Videos).
• A request’s details can be deleted after 90 days: /v1/generation then returns 404. Daily totals remain in the “Usage” tab of the portal.

Errors#Link to the Errors section

Common errors (key, rate) are in Errors.

See also#Link to the See also section

• Costs and wallet: holds, settlement, top-ups and invoices.
• Limits: workspace and key spend caps.
• Authentication: the response headers, including X-Moon-Request-Id.