Models & providers

The model field is how you choose a provider. One Open Vertex Router key reaches models from OpenAI, Anthropic, Google and others, so adding a vendor is a string change rather than a new integration, contract or invoice. All models in the catalog are chat models, served through POST /v1/chat/completions.

Model ids

Model ids are the vendor’s own names, with no provider prefix:

Vendor	Examples
OpenAI	gpt-4.1, gpt-4.1-mini, gpt-4.1-nano
Anthropic	claude-haiku-4-5, claude-opus-4-6
Google	gemini-2.5-pro, gemini-2.5-flash
DeepSeek	deepseek-v4-pro, deepseek-v4-flash
Mistral	codestral-latest
Moonshot	kimi-k2.5
Z.ai	glm-4.5

A few ids do keep a slash because that is the model’s published name, for example openai/gpt-oss-120b and meta-llama/Llama-3.3-70B-Instruct-Turbo. Some open-weight models are available from more than one host; the host then appears as a suffix, as in openai/gpt-oss-120b-groq. Where a vendor publishes dated releases, both the dated id (claude-haiku-4-5-20251001) and the undated one (claude-haiku-4-5) are available.

Whatever the form, the id to send is exactly what GET /v1/models returns. Copy it as is. An id with a prefix the catalog does not use, such as openai/gpt-4.1-mini, is rejected with 400.

List the models your key can call

curl

curl https://<your-gateway>/v1/models \
  -H "Authorization: Bearer $OVR_API_KEY"

Python

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

JavaScript

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

The response is the OpenAI list shape; the ids are in data[].id. When a key is restricted to some models, the list only contains those. The Models page of the dashboard shows the full catalog. The catalog evolves as vendors release models, so treat GET /v1/models as the current list rather than this page.

Restrict a key to some models

When you create or edit a key on the API keys page, you can give it a list of allowed models. A request for any other model is refused with 403 and a permission_error, and GET /v1/models with that key only lists what it may call. This is a simple way to keep a service on the models you have validated, or to keep an experiment away from the expensive ones.

Errors you may see

Status	Cause
400	The id is not in the catalog. The message points you to /v1/models.
403	The id exists but this key is not allowed to call it.

See Errors for the envelope and the full list.