Implementation:Helicone Helicone Unified Provider Registry

Knowledge Sources	Helicone
Domains	Provider Integration, AI Gateway, Configuration
Last Updated	2026-02-14 06:32 GMT

Overview

A centralized registry of LLM provider API configurations that maps each supported provider to its base URL, authentication format, default endpoint template, body mapper, and required environment variables.

Description

The providers.ts module defines providerConfigs, a record mapping each Provider (OpenAI, Azure, Anthropic, Bedrock, Google Gemini, Google Vertex AI, OpenRouter, DeepSeek) to a ProviderConfig object. Each configuration includes:

• baseUrl: The provider's API base URL, potentially containing placeholder tokens like {REGION}, {ENDPOINT}, or {DEPLOYMENT}
• authHeaderConfig: The authentication header name and optional value prefix (e.g., Authorization: Bearer vs x-api-key)
• defaultMapper: The body mapper name for request/response transformation (e.g., "openai-chat", "anthropic-chat", "gemini-chat")
• defaultEndpoint: The API endpoint path template
• envVars: Required environment variable names for authentication
• defaultHeaders: Optional static headers (e.g., Anthropic's anthropic-version)

The module exports getProviderConfig for simple lookups and buildProviderUrl for constructing fully resolved URLs by replacing placeholder tokens ({REGION}, {PROJECT}, {LOCATION}, {ENDPOINT}, {DEPLOYMENT}, {modelString}) with provided values.

Usage

Use this module when constructing API requests to any supported LLM provider, particularly in the playground and unified model layer.

Code Reference

Source Location

• Repository: Helicone
• File: packages/cost/unified/providers.ts

Signature

export const providerConfigs: Record<Provider, ProviderConfig>

export function getProviderConfig(provider: Provider): ProviderConfig

export function buildProviderUrl(
  provider: Provider,
  modelString?: string,
  providerSettings?: {
    region?: string;
    project?: string;
    location?: string;
    endpoint?: string;
    deployment?: string;
  }
): string

Import

import {
  providerConfigs,
  getProviderConfig,
  buildProviderUrl,
} from "@helicone/cost/unified/providers";

I/O Contract

Provider	Base URL	Auth Header	Default Mapper	Default Endpoint
OPENAI	https://api.openai.com	Authorization: Bearer	openai-chat	/v1/chat/completions
AZURE	https://{ENDPOINT}/openai/deployments/{DEPLOYMENT}	api-key	openai-chat	/chat/completions?api-version=2024-10-21
ANTHROPIC	https://api.anthropic.com	x-api-key	anthropic-chat	/v1/messages
BEDROCK	https://bedrock-runtime.{REGION}.amazonaws.com	Authorization	anthropic-chat	/model/{modelString}/invoke
GOOGLE_GEMINI	https://generativelanguage.googleapis.com	x-goog-api-key	gemini-chat	/v1beta/models/{modelString}:generateContent
GOOGLE_VERTEXAI	https://{REGION}-aiplatform.googleapis.com	Authorization: Bearer	gemini-chat	/v1/projects/{PROJECT}/locations/{LOCATION}/publishers/google/models/{modelString}:generateContent
OPENROUTER	https://api.openrouter.ai	Authorization: Bearer	openai-chat	/api/v1/chat/completions
DEEPSEEK	https://api.deepseek.com	Authorization: Bearer	openai-chat	/chat/completions

Function	Parameters	Returns	Description
getProviderConfig	provider: Provider	ProviderConfig	Returns the full configuration for a provider
buildProviderUrl	provider: Provider, modelString?: string, providerSettings?: object	string	Constructs a fully resolved URL with all placeholders replaced

Usage Examples

import { getProviderConfig, buildProviderUrl } from "@helicone/cost/unified/providers";

// Get OpenAI config
const config = getProviderConfig("OPENAI");
console.log(config.baseUrl);          // "https://api.openai.com"
console.log(config.defaultEndpoint);  // "/v1/chat/completions"

// Build a Vertex AI URL with all placeholders resolved
const url = buildProviderUrl("GOOGLE_VERTEXAI", "gemini-pro", {
  region: "us-central1",
  project: "my-project",
  location: "us-central1",
});
// "https://us-central1-aiplatform.googleapis.com/v1/projects/my-project/locations/us-central1/publishers/google/models/gemini-pro:generateContent"

// Build an Azure URL
const azureUrl = buildProviderUrl("AZURE", undefined, {
  endpoint: "my-resource.openai.azure.com",
  deployment: "gpt-4-deployment",
});
// "https://my-resource.openai.azure.com/openai/deployments/gpt-4-deployment/chat/completions?api-version=2024-10-21"

Related Pages

• Principle:Helicone_Helicone_Unified_Model_Layer