Submit and poll
Submit a job, poll its status, fetch the result. History, per-job detail and downloads are all addressable.
Video AI
One account, one API key, one rate per second that does not change when you scale. Everything below runs today on Phosor.
Video is billed per frame at 16 frames a second; the per-second figures here are the published per-frame price multiplied out.
| Capability | Model | Resolution | Price | Unit |
|---|---|---|---|---|
| Text to video Few-step distilled. The cheapest way to iterate on an idea. | Wan 2.2 turbo | 480p · 720p · 1080p | $0.005 · $0.010 · $0.018 | per second |
| Text to video Full sampling schedule. Use it once the shot is locked. | Wan 2.2 standard | 480p · 720p · 1080p | $0.015 · $0.030 · $0.040 | per second |
| Image to video Drive motion from a still. Same price ladder as text to video. | Wan 2.2 (standard / turbo) | 480p → 1080p | $0.005 – $0.040 | per second |
| Speech to video Lip-sync a character to an audio track. | Wan 2.2 s2v | 480p · 512p · 720p | $0.015 · $0.021 · $0.030 | per second |
| Motion transfer Move a character with a reference performance. | Wan 2.2 animate | 480p · 512p · 720p | $0.020 · $0.028 · $0.040 | per second |
| Multi-reference Images, video and audio references in one pass, with sound generated alongside the picture. First 5 reference images free, $0.009 each after. | MiniMax H3 ref2va | 480p · 768p | $0.0054 · $0.0144 | per second |
| Long-form video An alternative base with different motion characteristics. | LTX | 480p · 512p · 720p | $0.013 · $0.014 · $0.019 | per second |
| Image generation Stills for thumbnails, keyframes and product shots. | Qwen Image · Flux2 · Z-turbo | — | $0.0025 – $0.015 | per image |
| Image editing Instruction-driven edits on an existing image. | Qwen Image Edit (+ turbo) | — | $0.003 · $0.005 | per image |
| Speech synthesis Up to 500 characters per request. | Qwen TTS | — | $0.00003 | per character |
| Custom LoRA Upload a dataset and train, or import .safetensors from a URL. Several LoRAs per request. Generation with a LoRA is 1.2x the base rate. | Train, or bring your own | — | $0.003 | per training step |
Read from the platform's public pricing endpoint on 2026-09-22. Generation with a LoRA applied is 1.2x the base rate.
Commerce imaging, priced per job rather than per second.
| Job | What it does | Price |
|---|---|---|
| Product suite, 2K | A full set of shots from one product photo | $0.70 – $10.00 |
| Product suite, 4K | The same, at print resolution | $1.00 – $20.00 |
| Clothing suite | Garment shot to catalogue set, 2K or 4K | $1.70 – $2.00 |
| Flat-lay to model | Put the garment on a person | $0.40 |
| Mannequin to model | Replace the dummy with a model | $0.40 – $0.70 |
| Background change | Re-place the subject in a new scene | $0.40 – $0.70 |
| Pose variation | Another pose from the same model | $0.40 |
| Image translation | Localise the text inside an image | $1.00 |
10 credits to the US dollar. Auto top-up and invoices are available on the account.
Submit a job, poll its status, fetch the result. History, per-job detail and downloads are all addressable.
Issue API keys per environment. Balance, per-job charges and invoices are readable from the account.
Upload datasets for LoRA training, import LoRA weights from a URL, and keep generated output in platform storage.