Pricing

Estimate the cost of a run before you submit it. The quote is a pure function of the run shape — no job is created, nothing is billed. For the full pricing model and rate tables, see the Pricing page.

POST /pricing/estimate

Returns an itemized quote. Pass either tokens directly, or train_steps (tokens are then derived as train_steps × seq_len).

// request
{
  "type": "train",
  "params_before": 70000000000,
  "params_after": 90000000000,
  "train_steps": 5000,
  "seq_len": 256
}

// response
{
  "total_usd": 557.6,
  "unit": "1M-tokens",
  "qty": 1.28,
  "line_items": [
    { "label": "training", "qty": 1280000,    "unit": "1M-tokens", "rate": 45.0, "amount": 57.6 },
    { "label": "growth",   "qty": 20000000000, "unit": "params",    "rate": 25.0, "amount": 500.0 }
  ]
}

Request fields

Field	Type	Notes
type	train | contract	Defaults to train. Contraction returns the flat rate.
params_before	int | null	Base model size. Used to compute the growth line.
params_after	int | null	Output (grown) size. Selects the token tier and the growth amount.
tokens	int | null	Training tokens, if known.
train_steps	int | null	Used when tokens is omitted: tokens = train_steps × seq_len.
seq_len	int	Sequence length for the derivation. Default 256.

Response fields

Field	Type	Notes
total_usd	number	What the run will be billed (after the per-job floor).
unit / qty	string / number	The primary meter for this run (1M-tokens, or job for contraction).
line_items[]	array	Itemized charges — label, qty, unit, rate, amount. Render directly.

The same line_items shape lands on the usage ledger when a run finishes, so an estimate and the final bill are directly comparable.