On this page

POST /v1/sellers/{id}/capacity

Informational capacity snapshot. Does not change the book.

Auth: X-Seller-API-Key.

See rate limits and errors. How operators choose the numbers is in the capacity listing walkthrough.

Request

field	type	notes
`model_id`	string	Model the snapshot applies to
`available_tokens`	number	Tokens the seller can serve now
`queued_tokens`	number	Tokens already committed

Response

seller_id, model_id, available_tokens, queued_tokens, reported_at.

Errors

unauthorized, forbidden, not_found. Status codes on Errors.

cURL

curl -s http://127.0.0.1:8080/v1/sellers/SELLER_ID/capacity \
  -H "X-Seller-API-Key: sk-seller" \
  -H "Content-Type: application/json" \
  -d '{"model_id":"gpt-4","available_tokens":500,"queued_tokens":0}'

Rust

use token_gateway_sdk::{RegisterSellerRequest, ReportCapacityRequest, SellerClient, SellerKey};
// {"model_id":"gpt-4","available_tokens":500,"queued_tokens":0}

let mut client = SellerClient::new("http://127.0.0.1:8080", None);
let registered = client
    .register(&RegisterSellerRequest {
        name: "acme".into(),
        endpoint_url: "http://127.0.0.1:9000/infer".into(),
        supported_models: vec!["gpt-4".into()],
        api_key: "sk-seller".into(),
    })
    .await?;
client.set_key(SellerKey(registered.api_key.clone()));
let _ = client
    .report_capacity(
        registered.seller_id,
        &ReportCapacityRequest {
            model_id: "gpt-4".into(),
            available_tokens: 500,
            queued_tokens: 0,
        },
    )
    .await?;

Python

# {"model_id":"gpt-4","available_tokens":500,"queued_tokens":0}
from token_gateway import SellerTokenGateway

gw = SellerTokenGateway("http://127.0.0.1:8080")
seller = gw.register_seller(name="acme", endpoint_url="http://127.0.0.1:9000/infer", supported_models=["gpt-4"], api_key="sk-seller")
gw.report_capacity(seller.seller_id, model_id="gpt-4", available_tokens=500, queued_tokens=0)

TypeScript

// {"model_id":"gpt-4","available_tokens":500,"queued_tokens":0}
import { SellerClient } from "@ai-token-gateway/sdk";

const seller = new SellerClient({ baseUrl: "http://127.0.0.1:8080" });
const registered = await seller.register({
  name: "acme",
  endpoint_url: "http://127.0.0.1:9000/infer",
  supported_models: ["gpt-4"],
  api_key: "sk-seller",
});
seller.setApiKey(registered.api_key);
await seller.reportCapacity(registered.seller_id, { model_id: "gpt-4", available_tokens: 500, queued_tokens: 0 });