Pricing API

Authentication

Send your key as a bearer token. Keys are issued on the page you land on after paying, and are shown once.

curl -H "Authorization: Bearer sb_live_…" https://self-bench.com/api/v1/models

Endpoints

GET /api/v1/models

Every model in the catalog with prices, context window, output limit, modalities and the source link.

GET /api/v1/models/<slug>

One model. Returns 404 with a JSON error if the slug is unknown.

POST /api/v1/estimate

Cost for a workload, priced across every model or a listed subset.

curl -X POST https://self-bench.com/api/v1/estimate \
  -H "Authorization: Bearer sb_live_…" \
  -H "Content-Type: application/json" \
  -d '{"inputTokens": 1500, "outputTokens": 400, "requestsPerMonth": 6000, "models": ["qwen3.7-plus"]}'

GET /api/v1/export.csv

The whole dataset as CSV, regenerated on every request from the same source as the site.

Response shape

{
  "ok": true,
  "generatedAt": "2026-10-15",
  "data": {
    "slug": "qwen3.7-plus",
    "name": "Qwen3.7-Plus",
    "vendor": "Alibaba Cloud (Qwen)",
    "inputPerMillion": 0.3,
    "outputPerMillion": 1.19,
    "contextWindow": 1000000,
    "sourceUrl": "https://help.aliyun.com/zh/model-studio/model-pricing"
  }
}

Prices are USD per million tokens at published list rates. Every response includes the date the dataset was last refreshed.

Limits and fair use

• 10,000 requests per month on the Pro plan, counted per key.
• Cache the catalog locally; it changes at most a few times a week and generatedAt tells you when.
• No scraping the HTML site — that is what the API is for.