API reference
List models
Every model the gateway routes, in the OpenAI list format that SDKs read, plus the fields a model picker or cost estimator needs: vendor, context length, modalities and prices.
https://api.oneapi.id/v1/models- Key header
Authorization: Bearerorx-api-key- Body
- None
- Cost
- Free, not billed
Request
curl https://api.oneapi.id/v1/models \
-H "Authorization: Bearer $ONEAPI_API_KEY"Response
{
"object": "list",
"data": [
{
"id": "deepseek-v4.1-flash",
"object": "model",
"created": 1790701782,
"owned_by": "deepseek",
"name": "DeepSeek: DeepSeek V4.1 Flash",
"description": "Latest model from DeepSeek",
"context_length": 1048576,
"input_modalities": ["text", "image"],
"output_modalities": ["text"],
"protocol": "openai_chat_completions",
"pricing": {
"prompt": "0.00000021",
"completion": "0.00000084",
"input_cache_read": "0.0000000042",
"discount": 0.3
}
}
]
}idstringThe value to send as model.
objectstringAlways model.
createdintegerUnix time the route was added to the gateway.
owned_bystringThe vendor the model belongs to, as a lowercase slug taken from the catalog: deepseek, anthropic. other when the catalog does not place it.
namestringDisplay name. Falls back to id.
descriptionstringOne-line description. Empty when none was written.
context_lengthintegerContext window in tokens. Omitted when unknown.
max_completion_tokensintegerLongest output the model will produce, when one is set. Omitted otherwise.
input_modalitiesstring[]text, plus image when the model accepts images (vision).
output_modalitiesstring[]Always text.
protocolstringThe request shape the model answers on: openai_chat_completions, openai_responses, anthropic_messages or systemone. A systemone model only answers POST /v1/systemone.
pricingobjectWhat a request is billed at, described below.
Pricing
Prices are US dollars per single token, as decimal strings, the unit OpenRouter uses: "0.0000003" is $0.30 per million tokens. They are strings so no rounding creeps in; multiply by your token count. To show a per-million price, multiply by 1,000,000.
promptstringInput tokens.
completionstringOutput tokens.
input_cache_readstringInput tokens served from the provider's prompt cache.
input_cache_writestringTokens written to the prompt cache. Present only for models that charge for it.
discountnumberFraction off the vendor's published input price, from 0 to 1: 0.3 means 30% cheaper. 0 when there is nothing to compare against.
The list is the same for every key. A key limited to certain models still sees all of them, and gets 403 unknown_model when it calls one outside its list.
To check what your key can still spend, call GET /v1/me. To browse prices in a page, see Models or the model catalog.