Pricing
Every call is metered in credits. Media models are charged per call; chat models are settled by the token count the upstream reports. Prices below are generated from the ModelFold catalog, so this page can never drift from the API.
How credits are calculated
- Media (image / video / audio):
credits = ceil(upstream cost × 1.25), where per-second upstream prices are normalised to a 5-second clip. - Chat:
credits = ceil((input tokens × input price + output tokens × output price) / 1,000,000). Requests are pre-authorized, then settled on the real usage the upstream returns. - Refunds: if the upstream call fails, the charge is reversed automatically.
Video Generation
Image Generation
Chat Models
Audio & Speech
| Model | Credits / call | Upstream unit |
|---|---|---|
elevenlabs/text-to-dialogue-v3 | 8 | — |
elevenlabs/text-to-speech-multilingual-v2 | 15 | per 1000 characters |
elevenlabs/text-to-speech-turbo-2-5 | 8 | per 1000 characters |
gemini-omni-audio | 8 | — |
google/gemini-2-5-pro-tts | 8 | — |
google/gemini-3-1-flash-tts | 8 | — |
google/gemini-3-8-flash-lite-tts | 8 | — |
google/gemini-3-8-flash-tts | 8 | — |
infinitalk/from-audio | 8 | — |
wan/2-2-a14b-speech-to-video-turbo | 8 | — |
Prices marked
defaultin the catalog come from an upstream table that does not cover the model; they are modality starting prices and are pending manual review.