Ten placements free. Then $1 a month each.

Usage is metered per workspace and billed monthly against the three meters. Add a card when a workspace passes its free allowances and set the cap at the same time; until then there is nothing to pay. Pilots and quotes are invoiced.

Because they are not what Assetlib does. A workspace with two people and five agents publishes more artwork than one with ten people, and a seat price would undercount the first and tax the second. The bill follows the work: placements managed, bytes stored, bytes delivered. That is also why there is no "Pro" tier to unlock; every workspace has every feature.

Linear. Linear's MCP makes agents file more issues into the place Linear already charges for, so it costs Linear little and adds value; Figma's MCP hands design context to coding agents, which is the job a paid Dev seat used to do, so Figma caps it. An agent reading an Assetlib workspace connects more placements and prepares better drafts, which is the thing we bill. Reading is free; writes use the same meters as a person; only inference we pay for would ever be metered separately.

A named image slot in your code, with a size contract and a bundled fallback, that appears in an app's current published release: a promo card, a seasonal illustration, an onboarding frame, an empty state. Icons, vectors, and the images in a catalog are not placements, so an app with eight hundred imagesets is not eight hundred placements; the audit proposes at most three to start. The first ten in a workspace are free, the next ninety are $1 each, the next four hundred $0.50, and past five hundred $0.25 or a quote. Drafts and archived placements are not counted, and neither are the Roam, Daylight, and Still sample workspaces.

A placement is one slot however many variants it carries: light and dark, locales, segments, experiment arms. None of them is a second placement. What an experiment costs is what it delivers. Each device downloads only its assigned arm, so a two-arm test on 300,000 devices is about 300,000 downloads to start and half that again when the winner rolls out, roughly 54 GB, or about $9 at the delivery rate, plus a few cents of storage for the extra artwork. Assignment comes from the experiment or flag tool you already use; those connectors (LaunchDarkly, Amplitude Experiment, OpenFeature) are planned, and the rule holds when they ship.

Multiply devices by artwork changes. Each change sends about 120 KB to each active device that opens the app, so 100,000 devices and one change is about 12 GB, or $2.40 past the free 10 GB. An app with 300,000 devices changing artwork six times a month pays about $41 in delivery on top of its placements. Below about 50,000 devices, delivery is usually free.

Set the cap. It is a hard monthly maximum on the metered charges, and the console shows the three meters against it. Delivery to installed apps continues at the cap; what pauses is new work in the console. Most teams know their bill to the dollar from the placement count alone, because storage and delivery are small next to it until an app has a large audience.

Two guided working sessions with us, one to choose and connect the first placements with their fallbacks and one to run a publish and a restore on your installed build, plus $100 of metered usage and your per-unit rates held for twelve months. It costs $399, once, by invoice. It does not include app development, new SDK features, migration of a whole image library, or a production service level; the SDKs are developer previews.

Hosting bytes is not what you are paying for; at these rates the bytes are nearly free. The placement price pays for the release workflow around them: placement contracts, compatibility checks, signed releases, history, restore, and the bundled fallbacks that keep an app working when delivery fails. If an image CDN already covers your workflow, keep it.