Pricing model

Socaity uses three billing modes. Which one applies depends on the service you call and how it runs. You always pay only for what runs — never for waiting, cold starts, or failures.

Mode 1 — MaaS · pay-per-use

You call the platform, a GPU spins up, runs your inference, returns the result, and spins down. You are charged per unit of work, where the unit is the one shown on the service’s own page — it can be per millisecond, per second, per image, or per 1k tokens, depending on the service. Idle time, queue time, and failed runs are not billed. Like a taxi meter — it only runs while you are moving.

Mode 2 — Dedicated hosting

You reserve a GPU permanently. We rent it from a provider (RunPod today; Scaleway and Azure coming soon) and pass the rental cost on. The GPU stays on whether you send requests or not. You pay by the hour. Like renting an apartment — rent is due whether you sleep there or not.

Mode 3 — SPAINE workflows

For agentic workflows, billing is outcome-based: a percentage fee per finalised deliverable, not per token or per call. You pay when the workflow produces something you wanted. Like a sales commission — only when the deal closes.

• Cold starts — the seconds spent spinning up a GPU on the first call after idle.
• Queue latency — the time your job spent waiting for a free worker.
• Failed executions — if Socaity infrastructure fails the request, you are not charged.
• Idle services — serverless services scale to zero. Nothing running, nothing billed.

A subscription is optional. It layers on top of the four modes and unlocks higher quotas, priority queueing, environment cloning, and enhanced privacy.

Your usage breakdown and remaining credit are in the billing dashboard. See Platform → Billing for details.