Pricing

Prepaid credit. Published rates.

You buy credit once and it is spent on tokens at the provider's own published rates. There is no subscription, no seat and no minimum.

You pay $5.00
You get in usage $10.00

How the multiplier works: your credit is spent against the per-token rates below. A 2× multiplier means a $5.00 payment covers $10.00 of metered usage — no rate changes, no hidden spread.

Packages

Pick the amount of credit you need.

Starter
$5.00one-off

Buys $10.00 of usage at published rates.


  • 3 models included
  • DeepSeek v4.1 Flash
  • Meta Muse Spark 1.3 Contributor
  • Meta Muse Spark 1.2 Contributor
  • Unused credit does not expire on a cycle
  • Own API keys, create or revoke any time
  • Per-token usage metering in your console
Get Starter
Per-token rates

What each token costs.

USD per 1 million tokens, read across as input / cached input / output. Cached input is billed only for prompt tokens the provider served from cache.

Model Input Cached input Output Peak input Cache saving
Qwen3.6 Plus Qwen · Qwen/Qwen3.6-Plus $0.50 $0.10 $3.00 same
Qwen3.7 Flash Qwen · Qwen/Qwen3.7-Flash $0.03 $0.006 $0.13 same
Qwen3.8 Flash Qwen · Qwen/Qwen3.8-Flash $0.16 $0.016 $0.47 same 10×
DeepSeek v4.1 Flash DeepSeek · deepseek/deepseek-v4.1-flash $0.15 $0.003 $0.60 $0.30 50×
Meta Muse Spark 1.2 Contributor Meta · meta/muse-spark-1.2-contributor $0.10 $0.002 $0.20 same 50×
Meta Muse Spark 1.3 Contributor Meta · meta/muse-spark-1.3-contributor $0.10 $0.002 $0.20 same 50×
Z.ai GLM 5.3 Flash Z.ai · z-ai/glm-5.3-flash $0.15 $0.03 $0.50 same
DeepSeek v4 Flash Vision Exp DeepSeek · openrouter/deepseek/deepseek-v4-flash-vision-exp $0.15 $0.003 $0.60 same 50×
Rates are the upstream providers' published per-1M-token prices and are charged against your prepaid credit, which is why a payment covers 2× its face value in usage.
Peak hours. Where a model shows a higher peak input rate, that rate applies during the provider's busy periods — for DeepSeek v4.1 Flash, 01:00–04:00 and 06:00–10:00 UTC, Monday to Friday. It is set from each request's own timestamp and shown on that request in your usage log; it is never applied retroactively.
Mechanics

How the money actually moves.

How billing works

01Metered per token. Prompt tokens, cached prompt tokens and completion tokens are counted separately and priced at the rates above.
02Deducted from credit. Each request reduces your balance. There is no invoice to pay unless you ask for one.
03Hard stop, no surprise. When the balance is exhausted, requests return 429. We do not auto-charge a card on file.
04Auditable. Your console lists the balance and the requests that consumed it, with token counts and cost.

Included with every plan

  • Full context windows and streaming
  • Tool / function calling and JSON mode
  • Own keys scoped to your models
  • Prompt caching billed at the cached rate
  • Unused credit carries, it does not expire on a cycle
Questions

Billing, limits, and what is not charged.

Is there a monthly fee?

No. Credit is prepaid and spent as you use it. If you stop using the API you simply stop spending — there is nothing to cancel.

What happens if I go over?

Nothing breaks and no card is charged. Once the balance is spent, the API returns 429 with a budget message until you add credit. Your keys and configuration stay exactly as they are.

Why is cached input so much cheaper?

Upstream providers charge a reduced rate for prompt tokens they can reuse from a previous request. We pass that straight through — on DeepSeek v4.1 Flash it is 50× cheaper than standard input, which matters a lot for agent loops and stable system prompts.

Do you charge a markup on top of the rates?

The per-token rates are the providers' published prices. The discount you receive is the credit multiplier on your payment, shown above — not a quiet spread added to every token.

Are failed requests billed?

No. A request that errors — a rejected model, an invalid key, an upstream failure — is not billed. Only requests that return a completion are metered, and the usage log shows exactly which ones were.

Is there a minimum spend or lock-in?

No. There is no minimum, no contract and no monthly fee. You buy credit and spend it at your own pace; if you stop using the API you simply stop spending.

Are streaming responses billed differently?

No. Streaming and non-streaming requests are metered the same way, from the same token counts, at the same rates.

Do I need to talk to sales to get started?

No. There is no sales conversation, no approval step and no waitlist. Create an account, add credit and call the API.

Can I get an invoice?

Yes. Every purchase produces an invoice in your console, and administrator-marked payments are recorded the same way as card payments.

What about limits?

Traffic is shared across a pool of upstream capacity, so very high burst rates can be rate-limited at peak. Steady production traffic is unaffected. See the limits section of the docs.

Ready when you are.

Create an account and a key in a couple of minutes.