Models & renderers
The short list we actually run.
Every endpoint below is registered in the worker, priced from a rate we read on the provider's own page, and shown with its real availability. There is no grid of models we cannot call: a longer list would be a longer lie.
Fidelity renderer
FFmpeg · 0 credits
Ratios
16:9 · 9:16 · 1:1 · 4:5
Cinematic generation
Queued, not yet run
AI video
3 endpoints registeredPer-scene image-to-video, each scene a separate paid call with its cost shown before it is queued. Registered means the worker holds an adapter for the endpoint; the cinematic path that ships today is local motion, and no paid generation has run under supervision.
- Registered
Hailuo 2.3 Fast
fal.ai · 768p
- Price
- $0.19 per 6 s · $0.32 per 10 s
- Credits
- 4 / 7
The default endpoint for a paid cinematic scene if that tier is approved. 768p is upscaled at assembly and never presented as native 1080p.
- Registered
MiniMax H3 Max Turbo
fal.ai · 1080p
- Price
- $0.08 per second
- Credits
- 10 per 6 s
Native 1080p. Budgeted at the list price, not a promotional rate.
- Supervised only
Seedance 2.5
OneFast · 720p
- Price
- ≈ $0.12 per second
- Credits
- 3 per second
Second video provider. One supervised submit completed in 141 s; still disabled for customer traffic while its billing is reconciled.
AI image
1 endpoint registeredUpscaling only. Nothing here invents a room or a piece of furniture.
- Registered
Topaz Precision
fal.ai · up to 4×
- Price
- $0.08 per 24 megapixel block
- Credits
- 2 per block
Non-generative upscale. Adds no pixel the photograph did not contain.
What is not on this page
Virtual staging, an AI presenter, and every generator we have no adapter, no verified price and no supervised run for. They are tracked in public on the feature list with the milestone that owns them, and they will appear here the day they can be called.
A Fidelity film animates your own photographs with deterministic camera moves and spends no credits at all.