LLM
List LLM models and prices via API
GET /llm/models returns, in the format of OpenAI's GET /v1/models, the models active and enabled for your account, from OpenAI, Anthropic and Meta (Muse), with price, context window, output limit and support for tools and vision. Each id is what goes in model in calls.
The request
GET /api/llm/models Authorization: Bearer <merchant token>
What you can do
- Find out which models your account can call, from OpenAI, Anthropic and Meta.
- See each model's input and output price per 1 million tokens.
- See context_window and max_output_tokens to size the call.
- Know which models have supports_tools and supports_vision before sending tools or images.
- Build a model picker in your product from the list.
Rules and limits
- The list only carries models that are active and enabled for the account, and only from providers where you registered your key in /merchants/credentials/{provider}.
- The id is always provider/model and is the value of model and models in POST /llm/chat/completions.
- Price is in dollars per 1 million tokens. Some models have pricing.long_context: above above_input_tokens input tokens, the whole call, input and output, is charged at those prices.
- A call's max_tokens cannot exceed the model's max_output_tokens or the account cap.
- Only the merchant Bearer is needed, no x-customer-token.
Main fields
| Field | Type | Required | Description |
|---|---|---|---|
| id | string | Yes | The model value in calls, always provider/model, such as anthropic/claude-sonnet-5-5. |
| owned_by | string | Yes | The provider: openai, anthropic or meta. |
| context_window | integer | Yes | Input and output tokens that fit in one call. |
| max_output_tokens | integer | Yes | The largest max_tokens the model accepts. |
| supports_tools | boolean | Yes | Whether the model handles tools and returns tool_calls. |
| supports_vision | boolean | Yes | Whether the model accepts image_url in messages. |
| pricing.input_per_mtok | number | Yes | Input price in dollars per 1 million tokens. |
| pricing.long_context | object | Yes | The long-prompt tier, when the provider charges more above an input size. null when there is a single price. |
Who it is for
Model picker in your SaaS
Show users the available models with each one's cost, reading straight from GET /llm/models.
Cost estimates
Compute the expected cost of a task from the input and output prices.
Agents that pick the model
An agent reads the list and picks the cheapest model with supports_tools for the task.
Frequently asked questions
Why is a model missing from the list?
The list only carries models that are active, enabled for your account and from providers where you registered your key in /merchants/credentials/{provider}.
Who bills the usage?
The provider, on your key. Connect records the tokens and cost of each call, at the GET /llm/models price at the time.
What is pricing.long_context?
A long-prompt tier: when the input exceeds above_input_tokens, the whole call is charged at that tier's prices. It is null on models with a single price.
See Connect applied to your product.
Try it for 15 days with no card, or walk our team through your use case.
