Studio
Spend hours. Every job shows what it cost.
example
exampleYour gallery · kept on this device
Size
Quality
Count
Cost: 0.003 h
Gemini 2.5 Flash Image. A failed render never deducts hours.
Your private endpoint
Not running
Stops automatically when the time is up or your hours hit zero. Extend any time.
Indicative rate; the exact quote appears before you confirm.
Redeem hours
Hours are a claim on real compute. Pick a workload, a verified NVIDIA card consumes the hours, and the receipt is marked onchain. Priced before you send anything.
No file yet.
Nothing is consumed until a provider accepts and the result verifies. Images run immediately on the Images panel. Consumed receipts are marked onchain when the factory API is live.
Your redemptions
| When | Job | Proof · input hash | Quote | Status |
|---|
Last 7 days
| When | Job | Model | Hours |
|---|
Receipts on this device. Sign in with your wallet to see the factory's copy.
Agent keys
Give an AI agent or a script its own key. It spends your hours per call through an OpenAI-compatible endpoint, with the same receipts you see here.
Use it like any OpenAI-compatible API
Or let the agent pay per call, no key at all
Keys and pay-per-call open with the factory API. Until then, agents can use your own OpenRouter key directly.