Draft
~$0.004 per asset
Blockouts, iteration, and volume. Self-hosted open models on RunPod keep the cost near zero.
- TRELLIS.2 (MIT, self-hosted) ~$0.004
- SF3D <$0.001
- Sloyd parametric ~$0
Private beta
One agent that routes every 3D asset to the right model
Yuliia is a production copilot for game developers. She sends each asset request to the cheapest AI model that clears your style bar. Across the market, per-asset costs span 100x, from $0.004 (self-hosted TRELLIS.2) to $1.60 (Rodin at max quality).
Meshy · Tripo · Rodin · Hunyuan3D · TRELLIS.2 · SF3D
Pipeline
How Yuliia works
A seven-stage production pipeline, from style interview to engine export. From where you sit, it looks like four steps.
-
Set your style bar
A short style interview and a concept image lock the look. Every asset after that is measured against it.
-
Route to the right model
Yuliia picks the cheapest model that clears your bar: Meshy, Tripo, Rodin, Hunyuan3D, or self-hosted open models on RunPod serverless GPUs.
-
Make it production-ready
Retopology, UV unwrapping, and PBR texturing. Characters get rigging and animation through Meshy or Tripo, the only rigging APIs.
-
Check and export
A QC render check catches misses before you see them. Exports land as GLB or FBX, ready for Unity, Unreal, or Roblox Studio.
Routing
Three lanes
Not every asset deserves the same spend. Yuliia routes each request into one of three lanes based on what it is for. The per-asset costs below are real numbers, not estimates.
Standard
$0.05–$0.50 per asset
Textured, style-matched production assets. Meshy leads for stylized and hand-painted looks. Hunyuan3D self-hosting is geo-routed around its EU/UK/South Korea license exclusion.
- Hunyuan3D 2.1 (RunPod) ~$0.05
- Tripo ~$0.30
- Meshy ~$0.50
Hero
$0.40–$1.60 per asset
Flagship assets where quality is the whole point. Rodin Gen-2.5 via fal.ai, with quad topology and an optional HighPack pass.
- Rodin Gen-2.5 $0.40
- + HighPack +$0.80
The agent
Meet Yuliia
Yuliia is the producer on your art team who never loses track of a spec. She interviews you once about your style, then holds every generated asset to that answer.
She is opinionated about spend, honest about model quality, and quietly reroutes work when a cheaper model can clear the same bar.
Pricing
Plans for every stage
Credits are the unit of work: 1 credit = $0.05 retail. A draft asset costs 1 credit, a standard textured asset 10, a hero asset 25, and a rig with 2 animations 8.
Playground
$0 /mo
- 30 credits per month
- Draft lane access
- GLB/FBX export
Indie
$19 /mo
- 300 credits per month
- All three lanes
- GLB/FBX export
Recommended
Studio
$59 /mo
- 1,200 credits per month
- Style-lock across your project
- Priority generation queue
Team
$199 /mo
- 5,000 credits per month
- 5 seats
- API access
1 credit = $0.05 · draft = 1 cr · standard = 10 cr · hero = 25 cr · rig + 2 animations = 8 cr
Evidence
Routing you can audit
Yuliia's routing decisions rest on measurements, not marketing. Here is what that means in practice.
Monthly benchmarks
We benchmark every supported model each month on the same eval set. When a model improves or a price moves, the routing table moves with it.
Private beta, plainly
Yuliia is in private beta. This page collects a waitlist, nothing more. We invite in batches as capacity allows.
No invented numbers
Every figure on this page is a real per-asset cost or a published price. No testimonials we did not receive, no user counts we do not have.
FAQ
Common questions
Which game engines are supported?
Assets export as GLB or FBX with engine-ready presets for Unity, Unreal, and Roblox Studio. If your engine imports either format, it works with Yuliia.
Who owns the assets I generate?
You do. Assets you generate are yours to use in commercial projects. Where an underlying model carries a regional license restriction — Hunyuan3D's self-host license excludes the EU, UK, and South Korea — Yuliia geo-routes those requests to a compliant model automatically.
How does routing actually save money?
Per-asset costs across 3D generation models span roughly 100x, from $0.004 to $1.60. Most assets in a game do not need the most expensive model. Yuliia's eval-driven routing sends each request to the cheapest model that clears your style bar, so hero spend goes only where it earns its keep.
What happens when a generation is bad?
Every asset passes a QC render check before delivery. When a generation misses your style bar, Yuliia regenerates or reroutes to a stronger model before it reaches you. A failed attempt is her problem to fix, not yours.
When does Yuliia launch?
Yuliia is in private beta now, and we invite from the waitlist in batches. We are not publishing a launch date until we can stand behind one. Joining the waitlist is the reliable way to hear first.
Waitlist
Join the private beta waitlist
We invite in batches, in order. Your email is used for the invite and nothing else.