Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.ai-stats.phaseo.app/llms.txt

Use this file to discover all available pages before exploring further.

Use this recipe when you have not picked a video model yet and want a repeatable way to choose one before you submit paid jobs.

​
Goal

  • Find active public video models.
  • Filter by the input and output shape you need.
  • Check provider coverage and pricing before submitting a job.

​
1. Fetch the video model catalogue

Use the dedicated video models endpoint instead of hard-coding a provider-specific list. 
curl https://api.phaseo.app/v1/videos/models \
  -H "Authorization: Bearer YOUR_API_KEY"
Each item can include capability metadata such as:
  • model
  • name
  • status
  • input_types
  • output_types
  • supported_params
  • providers
  • pricing

​
2. Narrow the list by workflow shape

Before choosing the model, write down the minimum requirements for the workflow. Examples:
  • text-to-video only
  • image-to-video or first-frame input
  • audio generation required
  • duration and aspect-ratio controls
  • one specific provider or region
That makes it easier to reject models that technically work but do not match the product path you are building.

​
3. Check the parameters you actually need

Do not select the model only on brand or leaderboard familiarity. Check whether the model surface exposes the controls your request will rely on. Common examples:
  • duration
  • aspect_ratio
  • resolution
  • seed
  • generate_audio
  • input_references
  • provider-specific settings through provider_params

​
4. Check provider and pricing coverage

For production use, also check:
  • which providers expose the model
  • whether the model is active publicly
  • whether pricing is fixed, per-second, or provider-dependent
That matters because operationally you are selecting both a model and the provider routes that can actually serve it.

​
5. Save the chosen model in one config layer

After you decide, store the chosen model in one place:
  • a preset
  • one config file
  • one environment-owned mapping
Avoid copying model ids across multiple workers or route handlers unless the workflow genuinely needs per-caller variation.
Last modified on May 19, 2026