Models

The model catalog is served by the gateway in the OpenAI list shape with additional fields for pricing, discounts, capabilities and provider coverage.

GET/v1/modelsfiltered by your key's allowed_models / allowed_modalities

GET/v1/models/{id}

Model ids

Model ids are always vendor/name, lower-case, e.g. openai/gpt-4o-mini, anthropic/claude-3-5-haiku, meta/llama-3.1-70b-instruct. The vendor is the model's author, not the provider that serves it: a Llama model may be routed to any of several hosting providers, and the one actually used is returned in the provider field and X-LLM-Provider header of each response.

List models

bash

curl "https://api.smartapihub.com/v1/models" -H "Authorization: Bearer $LLM_API_KEY"

json

{
  "object": "list",
  "data": [
    {
      "id": "openai/gpt-4o-mini",
      "object": "model",
      "created": 0,
      "owned_by": "openai",
      "display_name": "GPT-4o mini",
      "modality": "text",
      "input_modalities": ["text", "image"],
      "output_modalities": ["text"],
      "context_window": 128000,
      "max_output_tokens": 16384,
      "pricing": { "input_per_mtok": 150000, "output_per_mtok": 600000, "currency": "USD", "unit": "micro" },
      "discount_bps": 0,
      "capabilities": { "streaming": true, "tools": true, "vision": true },
      "supported_parameters": ["temperature", "top_p", "max_tokens", "tools", "response_format", "seed"],
      "providers": 2
    }
  ]
}

Fields

Field	Type	Description
`id`	string	`vendor/name` — pass this as `model` in requests.
`owned_by`	string	The vendor (author) of the model.
`display_name`	string	Human-readable name for UIs.
`modality`	string	Primary output: `text`, `image`, `audio`, `video`, `music`, `embedding`. Decides which endpoint accepts the model.
`input_modalities` / `output_modalities`	string[]	What the model accepts and produces (e.g. `["text","image"]` for vision models).
`context_window`	int \| null	Maximum total tokens (prompt + completion).
`max_output_tokens`	int \| null	Maximum completion tokens.
`pricing`	object	Customer list price in micro-USD, with `currency: "USD"` and `unit: "micro"`. See below.
`discount_bps`	int	Discount applied to your organization/key for this model, in basis points (100 bps = 1 %).
`capabilities`	object	Boolean flags such as `streaming`, `tools`, `vision`, `json_mode`. Only capabilities supported by at least one enabled provider are `true`.
`supported_parameters`	string[]	Request parameters that will be honoured; others are ignored or rejected.
`providers`	int	Number of enabled providers that can serve this model — more providers means more fallback capacity.

Pricing fields and how to convert

All prices are integers in micro-USD (1,000,000 micro-USD = $1.00). Keys depend on the modality:

Key	Unit	Applies to
`input_per_mtok`	micro-USD per 1,000,000 input tokens	text, embedding
`output_per_mtok`	micro-USD per 1,000,000 output tokens	text
`cached_input_per_mtok`	micro-USD per 1,000,000 cached input tokens	text (when the provider reports cached tokens)
`per_request`	micro-USD per request	any
`per_image`	micro-USD per generated image	image
`per_audio_second`	micro-USD per second of audio	speech, transcription
`per_video_second` / `per_music_second`	micro-USD per second of generated media	video, music
`per_character`	micro-USD per input character	some speech models

To convert to dollars divide by 1,000,000. input_per_mtok: 150000 is $0.15 per million tokens; a 2,000-token prompt costs 2000 × 150000 / 1_000_000 = 300 micro-USD = $0.0003.

typescript

const usd = (micro: number) => micro / 1_000_000;
const promptCostMicro = Math.round((promptTokens * pricing.input_per_mtok) / 1_000_000);

The price shown is the list price for your organization. Your discount_bps is applied on top; the amount actually charged is returned per request in usage.x_llm_cost_micro. See Pricing for the full formula.

Get one model

bash

curl "https://api.smartapihub.com/v1/models/openai%2Fgpt-4o-mini" -H "Authorization: Bearer $LLM_API_KEY"

Returns a single model object (same shape as list items). The slash in the id may be sent literally or percent-encoded. Unknown or disabled ids return 404 model_not_found; ids outside your key's scope return 403 model_not_allowed.

created is 0 for catalog entries — the platform does not track vendor release dates. Use display_name and capabilities to present models in a UI.