Pricing
Pay for output, not seats.
One unit. One price. No mental math. Pricing live for invited workspaces · Public from $49/mo Q3 2026.
Solo
$49/mo
For one operator running one brand.
- 1 workspace · 1 brand LoRA
- 500 hero images / month
- 10 hero videos / month
- 1 distribution channel (Shopify or Meta or Klaviyo)
- Dual-judge QA
- C2PA signing
- Email support
Agency
$199/mo
Multi-client brand portfolios.
- 5 workspaces · 5 brand LoRAs
- 2,500 hero images / month
- 50 hero videos / month
- All distribution channels
- Voice clone (1 brand)
- Workspace-scoped audit log
- Priority email support
Scale
$499/mo
Volume operations.
- 15 workspaces · 15 brand LoRAs
- 10,000 hero images / month
- 200 hero videos / month
- Voice clones for every brand
- Priority GPU on Forge
- TikTok Shop distribution (when live)
- Dedicated Slack channel
Enterprise
Custom
Category-leader infrastructure.
- Unlimited workspaces
- SSO / SAML / SCIM
- DPA + IP indemnification passthrough
- Dedicated LoRA training capacity
- Multi-region deployment
- Audit log export to your SIEM
- Dedicated success engineer
Open books
What an output actually costs us.
Most AI tools obscure unit economics. We don't, because the moat isn't margin — it's the production line. Here's roughly what each call costs us:
| Output | Our cost |
|---|---|
| Hero image (Flux 1.1 Pro Ultra avg) | ~$0.08 |
| Batch image (Flux Schnell, RunPod) | ~$0.012 |
| Hero video (Veo 3, 5s) | ~$0.50 |
| Talking head 30s (HeyGen Rachel + voice clone) | ~$2.40 |
| Magnific 4K upscale | ~$0.04 |
| C2PA signing per asset | ~$0 |
Margin pays for the production line — engineers, GPU reserves, the dual-judge QA stack, the spend-cap monitoring that means no $5K Friday-night surprises.
Prices are upstream-provider averages observed across our last 30 days of dogfood traffic.
FAQ
The pricing questions we hear most.
What's a 'unit' — is one image equal to one video?
No — image and video are billed separately because they cost differently to produce. A hero image runs us ~$0.08 (Flux 1.1 Pro Ultra avg). A 30-second talking-head video runs us ~$2.40 (HeyGen Rachel + voice clone + post-prod). Your tier sets a cap per kind, not a single 'credit' bucket — so you don't burn your video budget on background images.
What happens if I exceed my monthly cap?
We pause generation and ping you. No surprise overages. You can add a single-month top-up, or upgrade tier — both take effect immediately, no proration drama.
Why is there an 'invited beta' phase?
Because we need to be in the dispatcher loop with the first 30–50 customers to make sure spend caps work, the LoRA training quality is consistent across verticals, and the dual-judge thresholds are calibrated. We invite cohorts of 10–15 every two weeks. Public pricing kicks in Q3 2026.
Do I keep my LoRA when I cancel?
Yes. Your LoRA, your reference set, your audit log — all of it exports as a portable bundle. We're a tool, not a hostage situation.
How does the IP indemnification on Enterprise work?
We pass through the indemnification we get from upstream providers (Vertex AI for Imagen / Veo / Gemini, Adobe for Firefly when used). Combined with the C2PA signing on every output and the training-data disclosure we publish, it gives your legal team the full provenance chain for any asset that ships.
What if I'm an agency with bigger ambitions?
Talk to us. The Enterprise tier exists precisely for agencies running 50+ client brands — bespoke pricing, dedicated GPU, your team's IP indemnification umbrella. Reply via /contact and we'll get back same day.