Skip to content

Models and prices

Read live from GET /v1/catalog, so this list is what you can call right now, at today's prices.

Loading the catalog...

A model's exact input

curl https://api.factorialcompute.com/v1/catalog/sam3-segment

returns input_schema (JSON Schema - what a client or a coding agent writes code from), input (each field in a sentence), an example that works as written, output (the kind of answer), and pricing.

How prices work

  • You pay per unit of the work - per image, per object, per second of video, per million tokens - never per GPU-hour. The unit is one you can count before you send, so POST /v1/quote is exact for everything but language models, where it is an upper bound set by max_tokens.
  • Language models are at their supplier's list price.
  • There is no subscription and no minimum: runs draw down your account's credit. GET /v1/billing shows the balance; GET /v1/usage what it went on.