Models

Two ways to find a model

Use the public Model Catalog at /pricing to compare capabilities and live prices before integrating. Use GET /v1/models with your API key to retrieve the model identifiers currently available to that key.

The API result is authoritative for access at request time. A model can appear in the public catalog but still be unavailable to a restricted key or group.

List available models

GET /v1/models

cURL

curl 'https://hippoapi.net/v1/models' \
+  -H "Authorization: Bearer $HIPPOAPI_API_KEY"

Python

models = client.models.list()
for model in models.data:
    print(model.id)

JavaScript

const models = await client.models.list()
for (const model of models.data) {
  console.log(model.id)
}

Use the exact identifier

Copy the id value without changing capitalization, punctuation, or version suffixes. Put that string in the model field of the request. A display name, vendor name, or marketing title is not necessarily a valid model identifier.

Keep model identifiers in application configuration rather than scattering them through source code. This makes controlled model changes easier to review and roll back.

const MODEL = process.env.HIPPOAPI_MODEL || 'glm-5.1'

Do not assume every parameter is supported

OpenAI-compatible describes the request and response protocol. Model behavior and supported features still differ. Test streaming, tool calling, structured output, image input, context length, and sampling parameters with the exact model you plan to deploy.

When switching models, test representative inputs and compare quality, latency, error rate, and total charged usage. Changing only the model string is technically simple, but it is still an application behavior change.