Sevren docsAccount & access

Holder benefits

Wallets that hold the Sevren token get free tokens on the paid models: a one-time free try and, with a holder plan, free tokens every day. Each model has its own allowances, read live from the API.

What holders get

Sign in with an Ethereum wallet. The benefits follow the wallet's share of the token's total supply. The shares and amounts below are read live; the operator can change them at any time.

  • Free try: from the share shown for the free try, free prompt and output tokens on each model that offers them, once per wallet.
  • OUTLAW and LEGEND plans: from each plan's share, free prompt and output tokens on each model the plan covers, every day. The account gets the best plan its share reaches.

Free tokens per model

Read live from trialTokens and freeTokens in GET /v1/catalog and the plan tiers of GET /auth/wallet/config. Prompt and output tokens are counted separately, and each model's allowance is independent of the others.

Loading the live holder benefits…

The free try

  • It belongs to the wallet and is given once. It never resets.
  • It is a budget of tokens on each model, not a number of requests: it pays for as many requests on that model as it covers, with no limit per request, until it runs out.
  • A request that needs more than what is left uses the rest of the free try; the plan's free tokens or credit pay the remainder.
  • GET /account/me reports it in trial: per model, the total, used and remaining prompt (input) and output tokens, and whether the account is eligible.

Daily plan tokens

  • An active plan gives its free tokens on each model it covers, every UTC day. On a model where the plan has none, requests are paid from the free try, then from credit.
  • They reset at 00:00 UTC (freeTokens.resetsAt). Unused tokens do not carry over.
  • A running request holds its share, so remaining can be lower than total − used until it ends.
  • GET /account/me reports them per model in freeTokens.models.

Order of spending

On a paid model, each request is paid in this order, prompt and output tokens separately:

  1. the free try's tokens on that model;
  2. the plan's free tokens of the day on that model;
  3. account credit, at the model's prices.

One request can be split across them, for example part of the prompt from the free try, the rest from the plan and the output from credit.

When the free tokens run out

  • When the free try or the day's plan tokens on a model are used up, requests on that model are billed from your credit at the model's prices. Other models keep their own allowances.
  • Anyone, holder or not, can add credit at any time with an on-chain payment from the account page and keep using every model.
  • With neither free tokens nor credit left, a request answers 402 insufficient_quota.
  • Plan tokens come back at 00:00 UTC. The free try does not come back.

How holdings are checked

The free try and the plans use the same checks: the share counts what the wallet held at both the latest and the finalized block, it is re-checked every few minutes, and the benefits pause while the last check is more than one hour old. POST /account/plan/refresh checks it on demand, up to 6 times per hour. The details are in Billing.

See also

  • Billing: credit, prices, holds and the order of spending.
  • Payments: how a payment becomes credit.
  • Account API: trial and freeTokens in GET /account/me.