Pay per render, not per seat.

One fresh render = one credit, regardless of size or format. Cache hits are free: the same template with the same inputs returns the stored file and bills nothing. Previews (in the editor, over the MCP, in the demo above) are always free.

A fresh render takes about a second. A cached repeat serves in roughly 19 ms. Each render node sustains around 27 renders per second and nodes scale horizontally, so batch throughput is a provisioning question, not an architectural one.

Yes. Add lang=de to a render and every text layer is translated, or set it per field. The model runs on our own servers: no per-word vendor fees, your copy never leaves the press, and each translation is cached for 90 days so repeats are instant.

Upload a CSV in the dashboard or POST rows to /v1/render/batch. Each row becomes one image. Progress streams live, failed rows are listed individually, and when the run finishes you get an email and a batch.completed webhook containing every URL.

Everything you can: create templates, add and arrange elements, search stock, mark variables, preview for free, and render. The hosted MCP server exposes 26 tools and is listed in the official registry as pro.pixeldrive/pixeldrive. Free previews mean the agent iterates on real pixels before it spends a credit.

PNG (default), JPEG, WebP, AVIF and single-page PDF, with a qualityparameter for the lossy ones. Each format is cached separately, so switching formats doesn't invalidate what you've already rendered.

On our edge store, served from a stable URL you can hotlink anywhere: email, meta tags, your own CDN. Re-rendering the same inputs always resolves to the same file, so links never rot when your data doesn't change.

Cache hits keep serving free, so live traffic never breaks. Fresh renders pause until you top up or the month rolls over. Per-token rate limits are yours to set, so one runaway integration can't drain the account.