Model catalog - pre.dev

The catalog lists every model and provider the gateway serves, with pre.dev credit prices added. Reads are free, not rate-limited, and do not need credits.

Method	Path	Returns
GET	`/v1/models`	All models, each row with a `predev` price object
GET	`/v1/models/count`	Number of models
GET	`/v1/models/{author}/{slug}/endpoints`	The provider endpoints behind one model
GET	`/v1/embeddings/models`	Embedding models
GET	`/v1/images/models`	Image generation models
GET	`/v1/videos/models`	Video generation models
GET	`/v1/providers`	Providers

curl --fail-with-body -i https://api.pre.dev/v1/models \
  -H "Authorization: Bearer $PREDEV_API_KEY"

The `predev` object

Each model row carries its prices in credits (see pricing):

Field	Unit	Present when
`credits_per_m_input`	Credits per million input tokens	Always
`credits_per_m_output`	Credits per million output tokens	Always
`credits_per_m_cache_read`	Credits per million cached input tokens read	The model prices cache reads
`credits_per_m_cache_write`	Credits per million tokens written to cache	The model prices cache writes
`credits_per_image`	Credits per image input	The model prices image inputs
`credits_per_request`	Credits per request	The model has a per-request price

{
  "id": "anthropic/claude-sonnet-5",
  "name": "Anthropic: Claude Sonnet 5",
  "context_length": 1000000,
  "pricing": { "prompt": "0.000003", "completion": "0.000015", "input_cache_read": "0.0000003", "input_cache_write": "0.00000375" },
  "predev": {
    "credits_per_m_input": 42.86,
    "credits_per_m_output": 214.29,
    "credits_per_m_cache_read": 4.29,
    "credits_per_m_cache_write": 53.57
  }
}

The numbers above are illustrative; read the live row for current prices. The other fields in the row (name, context_length, dollar pricing, supported modalities and parameters, and so on) describe the model itself.

Freshness

x-predev-catalog-age on every catalog response is the number of seconds since the catalog last refreshed. The catalog refreshes every 15 minutes, so a newly released model can be callable before it appears in the list. Calling a model id that is not yet listed still works if the gateway serves it.

Endpoints for one model

curl --fail-with-body https://api.pre.dev/v1/models/deepseek/deepseek-v4.1-flash/endpoints \
  -H "Authorization: Bearer $PREDEV_API_KEY"

Use this with provider preferences to pin or exclude specific providers. See routing and fallbacks.