Skip to main content

​
Overview

The models endpoint returns a list of all available models that can be used with the Helicone AI Gateway. This endpoint is OpenAI-compatible and returns models in the same format as OpenAI’s /v1/models endpoint.

​
Authentication

This endpoint does not require authentication and can be called without an API key.

​
Endpoint

GET https://ai-gateway.helicone.ai/v1/models

​
Request

This endpoint accepts no query parameters and returns all available models. 
curl https://ai-gateway.helicone.ai/v1/models

​
Response Format

​
object
string
Always returns list to indicate this is a list response.
​
data
array
Array of model objects available through the gateway.

​
Example Response

{
  "object": "list",
  "data": [
    {
      "id": "gpt-4",
      "object": "model",
      "created": 1704067200,
      "owned_by": "openai"
    },
    {
      "id": "gpt-4-turbo",
      "object": "model",
      "created": 1704067200,
      "owned_by": "openai"
    },
    {
      "id": "gpt-3.5-turbo",
      "object": "model",
      "created": 1704067200,
      "owned_by": "openai"
    },
    {
      "id": "claude-3-opus-20240229",
      "object": "model",
      "created": 1704067200,
      "owned_by": "anthropic"
    },
    {
      "id": "claude-3-sonnet-20240229",
      "object": "model",
      "created": 1704067200,
      "owned_by": "anthropic"
    },
    {
      "id": "claude-3-haiku-20240307",
      "object": "model",
      "created": 1704067200,
      "owned_by": "anthropic"
    },
    {
      "id": "gemini-pro",
      "object": "model",
      "created": 1704067200,
      "owned_by": "google"
    },
    {
      "id": "gemini-1.5-pro",
      "object": "model",
      "created": 1704067200,
      "owned_by": "google"
    },
    {
      "id": "llama-3-70b",
      "object": "model",
      "created": 1704067200,
      "owned_by": "meta"
    },
    {
      "id": "mistral-large",
      "object": "model",
      "created": 1704067200,
      "owned_by": "mistral"
    }
  ]
}

​
Model Filtering

The endpoint automatically filters out models that:
  • Require explicit routing configuration
  • Are not available through the standard gateway endpoints
  • Have no registered endpoints in the cost registry
This ensures you only see models that are immediately usable with the chat completions endpoint.

​
Using Model IDs

The id field from each model object can be used directly in your chat completion requests: 
{
  "model": "gpt-4",
  "messages": [
    {"role": "user", "content": "Hello!"}
  ]
}

​
Model Support

The Helicone AI Gateway supports models from multiple providers:
  • OpenAI: GPT-4, GPT-3.5, and variants
  • Anthropic: Claude 3 family (Opus, Sonnet, Haiku)
  • Google: Gemini models
  • Meta: Llama models
  • Mistral: Mistral and Mixtral models
  • And many more: The registry is continuously updated with new models

​
Error Handling

​
error
object
Error information if the request fails.

​
Example Error Response

{
  "error": {
    "message": "Failed to fetch models from registry",
    "type": "internal_error"
  }
}

​
Implementation Details

The models list is generated from Helicone’s internal cost registry (@helicone-package/cost/models/registry), which:
  • Tracks pricing and capabilities for each model
  • Is automatically synced with provider model releases
  • Includes metadata like creation dates and ownership
  • Powers cost calculation and analytics in Helicone

​
Rate Limits

This endpoint has generous rate limits as it’s a read-only operation that doesn’t consume compute resources. It can be called frequently for model discovery without concerns about rate limiting.

​
Best Practices

  • Cache the response: The model list doesn’t change frequently, so cache it client-side to reduce API calls
  • Filter by provider: If you only use specific providers, filter the response by the owned_by field
  • Dynamic model selection: Use this endpoint to build dynamic model selection UIs
  • Validation: Validate user-provided model IDs against this list before making completion requests

​
Client-Side Filtering Example

// Get all OpenAI models
const openaiModels = models.data.filter(m => m.owned_by === 'openai');

// Get all Claude models
const claudeModels = models.data.filter(m => m.owned_by === 'anthropic');

// Check if a model exists
const hasGPT4 = models.data.some(m => m.id === 'gpt-4');

// Get the newest models (sort by created timestamp)
const newestModels = [...models.data]
  .sort((a, b) => b.created - a.created)
  .slice(0, 5);