Skip to content
tenvoralpha
GuidesBilling and receipts

Documentation · Reference

Billing and receipts

Your account holds prepaid credit, in dollars, and each request takes only what it used. A monthly plan adds an allowance of usage every week.

Your balance

Your balance is prepaid credit, in dollars. Card payments are not open yet. A new account starts with $0.25 of free credit.

Your balance is on the Billing page in the Console.

Plans

A plan gives your account an allowance of usage every week, measured at normal API prices. There are three plans, and each is billed monthly. Plans are paid for by card, so they open together with card payments.

The three plans
PlanPriceUsage each weekUsage each monthChat messages each week
Lite$6 a month$1.75$7.00250
Plus$12.50 a month$3.75$15.00600
Max$19 a month$5.75$23.00900

Your billing month is split into four equal parts, and each part starts with a fresh allowance. Use it at any time in the week: there is no five-hour limit.

The allowance covers every API key on your account and Work. Each plan also includes chat messages with Tenvor-35B in Chat & Work every week, and they do not use the allowance. Each one carries the most recent part of the conversation, about 6,000 words, and replies with up to about 750 words. After those, chat uses the allowance. Web searches use the allowance.

Usage takes the week's allowance first, then rollover credit, then credit you have bought.

When a week ends, half of the allowance you did not use is added to your account as rollover credit. Rollover credit expires 90 days after it is added and has no cash value.

The plans and their prices are also on Pricing.

What a request costs

You pay per token, for what goes in and what comes out. For each request you pay the larger of the tokens you used or the minimum per request. Rates and the minimum are on Pricing. They are not repeated here, so they cannot be out of date.

If a request fails on a node before any output is sent, Tenvor may retry it once on another node, if one is free. You are not charged for the attempt that failed. If a request fails on a node after output has started, you are not charged and it is not retried. A request that you cancel, or that stops at your own output limit, can be charged for its input and the output it generated.

When you are charged. If the model generates output for your request, you pay for it, whether or not it is an answer.

  • A reply with no answer is charged. Thinking is output. If it never closes, or closes before an empty answer, you pay for what was generated.
  • No output, no output charge. A request that generates nothing is not charged for output, but a Stop after the prompt has been read and before any token is still charged for the input and the minimum.
  • A stream is charged when it finishes or when you stop it. If you stop it, you pay for the input and for the output delivered up to that point.
  • A node that fails partway does not charge you. A stream that ends because its node failed is not charged. Only a stream that finishes, or one you stop, is billed.

Two limits of thinking are in the reference.

What is held while a request runs

When a request starts, Tenvor holds an amount for it. The Console shows it as Reserved. When the request finishes, Tenvor takes the real charge and releases the rest. What you can spend is your balance minus what is reserved.

The amount is the full input plus the most output the request allows. If you leave out max_tokens, Tenvor holds funds for a long default reply, so set it to the length you need. The Trust page gives the default.

If available balance cannot cover the reservation, the chat endpoint returns 402 insufficient_quota. Check Billing in the Console.

Read a receipt

Every request that completes has an itemised receipt. It is the billing object in the response to a request that does not stream. For a streamed request, fetch it at GET /v1/results/{id}, with the id from the X-Request-Id header. A streamed response itself ends with token usage.

A request that was cancelled has the amount charged, if any, and no itemised receipt. Failed requests have no receipt except paid request_output_limit failures, which include usage and billing.

A failed result that was proven free says so: its body carries billing.charged_micro_usd of 0, and no other receipt field.

The fields of a receipt, with the sample from the first request
full_input_tokens14Tokens you sent, at the full input rate.
cached_input_tokens0Input tokens billed at the cached rate, half the full input rate. It reads 0 on a first request and on any request that did not ask for reuse.
output_tokens9Tokens in the reply.
charged_micro_usd…What you are charged, in millionths of a dollar: 1,000,000 is $1. The rates are on Pricing.

Two more fields can be there. input_tokens_estimated says whether the input count is an estimate. billing.cache_status describes billing: it reads hit when cached input tokens were billed and miss when none were. tenvor_session_kv.status describes physical prefix reuse, what the node did with its saved state, so the two can differ. The rates the request was priced at are in the receipt too.

The Usage page in the Console lists every credit and charge on the account, and draws your spend for each day. Match a receipt to a charge by its amount and time.

A charge you do not expect

Check the Usage page first. If a request has no receipt, or the charge does not look right, write to support@tenvor.io. Give the time, the status and error code. If a person at Tenvor needs the X-Request-Id, they will ask for it. Leave out your key and the content of the request.

The service credits policy covers refunds and unspent balance.

Content checked . Live limits and prices can change.