Tunings

POST /v1/tunings · examples never stored

New tuning Step 1 of 2 · examples

Method

—

Examples · training tokens

—

GPU time (estimate)

—

Total (estimate)

—

Question	Type	Examples	Note

GPU time is passed through at our H100 rate: the quote is an estimate, you are billed the actual seconds. A rejected job is billed (the report is the deliverable); a failed job is refunded. Hosting is $10 per model per month after the first 30 days.

Held-out examples · before → after

Question	Type	Train / held out	Accuracy	ECE	NLL	T	Gate
No report yet.

Canary suites

—

Baseline

—

Use this model

Deleting removes the adapter and stops hosting. Your examples were never stored.

Model	Status	Method	Created	Cost	Actions
Loading…

Readout $10 / job · LoRA $10 / 1M training tokens ($25 min) · GPU at cost · hosting $10 / model / month after 30 days · tuned inference $0.40 / 1M Pricing →