Skip to content

08 · Cost, plans, and teams ​

The headline first: the app is free forever, Apache-2.0. The models cost money.

On top of that sits an optional managed service selling "you don't manage keys" plus cloud deploys. It isn't a prerequisite for using OpenDesign.

Free ​

Price$0, forever
LicenseApache-2.0, source on GitHub
ModelsBYOK — your key, billed to your provider account, never proxied
RunsLocally, macOS and Windows
IncludesThe full skill and system library, every agent adapter, artifacts in your own repo

For most individuals this is the tier. Especially if you already pay for Claude Code or Codex — design and coding draw on the same pool, so the marginal cost is near zero.

Managed tiers ​

You're buying convenience: no key management, a bundled model allowance, and cloud deploys.

TierMonthlyYearlyModel creditsCloud deploys
Plus$20$168$20 / mo3 / mo
Pro$100$720$120 / mo20 / mo
Max$200$1,176$300 / mo50 / mo

Deploys beyond the allowance are $2 each.

Note that Pro and Max include more credit than they cost ($100 buys $120; $200 buys $300). So if your usage is already at that level, the managed tier is slightly cheaper than buying API access yourself. At low usage, free plus your own key wins.

Teams ​

TierStandard / seat / moModel creditsMinimum seats
Team Basic$5none3
Team Plus$25$20 / seat / mo3
Team Pro$105$100 / seat / mo3
Team Max$205$200 / seat / mo3

Each has carried an intro price at various points (Team Pro at $73.50, for instance). Check the live page.

Team Basic is the interesting one: no model credits at all. It sells collaboration and nothing else — shared projects, shared design systems, shared plugins. That fits a team where everyone already has their own agent subscription and what's missing is a common set of brand assets and methods. It's often the most practical tier in the list.

Enterprise ​

Custom, covering shared team design systems, real-time collaboration, multi-user project and artifact editing, unified billing and usage, SSO / SAML, and priority support. Contact is support@open-design.ai.

Prices change

These numbers come from the machine-readable pricing page, open-design.ai/pricing.md. Promotional and annual pricing move often — check the live page before signing anything.

Containing the spend ​

Whatever tier you're on:

One: the front of the workflow is free. Steps 0 through 2 — purpose, style, structure — burn almost no generation credit. Thinking first is cheaper than regenerating later.

Two: don't regenerate to fix details. One word is an Edit, at effectively zero cost. Regenerating a page is the most expensive operation available.

Three: refine in narrow passes. Ask for seven dimensions at once and it rewrites everything, and you have to re-review all of it. One or two at a time keeps changes checkable.

Four: images and video dominate the bill. Generation and rendering there cost far more than text. Approve one image before batching.

Five: freeze stable work as skills. Skills remove rounds of clarification, which is where the long-term saving is.

Organising a team ​

A structure that works:

text
team-design/
├── design-systems/
│   ├── our-brand/DESIGN.md
│   ├── client-a/DESIGN.md
│   └── client-b/DESIGN.md
├── skills/
│   ├── monthly-report/SKILL.md
│   ├── feature-launch-deck/SKILL.md
│   └── client-proposal/SKILL.md
└── artifacts/
    └── 2026-09/

Three things live in the repo: brand systems, method (skills), and artifacts. Everyone brings their own agent subscription; what's shared is the repo.

This runs on the free tier with nothing but git. A Team plan buys the in-product collaboration experience — live co-editing, unified usage — not this structure.

A question that comes up constantly ​

"Can we skip the desktop app and just use it in the cloud?"

Not today. OpenDesign is local-first by design, and the minimum is a local daemon plus an agent. The project's own answer is blunt: a hosted sandbox is on the roadmap but not the priority, because artifacts in your repo beat documents in someone else's database.

Self-hosting is fine — the daemon runs anywhere Node 24 runs. Teams doing a shared deployment usually pin the daemon to a machine inside their network and point each developer's CLI at it.

An unofficial community guide · compiled from open-design.ai public material and hands-on notes · the official docs win where they differ