Pricing
Pay as you go with no monthly fee. The table shows each model's list price. Public requests add a 20% platform fee. Every request's cost can be looked up.
Ready to start?
Sign up, create an API key and call every model with the official OpenAI SDK.
Get startedPay as you go
No monthly fee or minimum spend. Usage is deducted from your balance in US dollars.
Hold, then settle
Before a request runs we hold its maximum possible cost, then charge the actual cost and release the rest. You are never charged more than the hold, and requests are rejected up front if your balance is too low.
Failures are free
Failed requests and failed output uploads are not charged; the hold is released in full.
All model prices
| Model | Input | Cached input | Output |
|---|---|---|---|
| $12.00 | - | $60.00 | |
DeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flash | $0.04152 | $0.012 | $1.20 |
| $0.90 | - | $4.50 | |
Muse Spark 1.3meta/muse-spark-1.3 | $1.50 | - | $5.10 |
| $12.00 | - | $60.00 |
Prices in USD per million tokens.
Billing units
Text
Per million tokens, with separate prices for input, cached input and output.
Image
Per image, priced by quality or resolution tier; models that return several images at once (such as Midjourney) are priced per request. GPT Image, FLUX.2 and MAI Image are billed by tokens (input tokens plus output image tokens), and output tokens grow with resolution and quality.
Video
Per second of requested duration, priced by resolution and whether audio is generated.
FAQ
Which price applies to me?+
This page and the pricing field from GET /v1/models are list prices, before the platform fee. Public accounts are charged list price × 1.20. Quotes and usage amounts are what you are charged, not the catalog price.
How do I see what a request cost?+
Every response includes usage.cost. You can also call GET /v1/generations/{id}, or open Usage in the dashboard for daily and per-request breakdowns.
Do price changes affect requests in progress?+
No. The price is locked when a request is accepted, so in-flight requests (including queued async jobs) settle at that price. New prices apply only to later requests.
Which currency do you bill in?+
All prices and charges are in US dollars, precise to 8 decimal places.


