Vercel AI SDK Integration

Helicone integrates seamlessly with the Vercel AI SDK, providing observability for streaming AI applications built with Next.js and React.

Integration Methods

Provider Configuration

Configure AI SDK providers to route through Helicone.

AI Gateway

Use the AI Gateway for unified access to all providers.

Custom Headers

Add Helicone headers to track sessions and custom properties.

Async Logging

Zero-latency logging for production apps.

Quick Start

Install dependencies

npm install ai @ai-sdk/openai @ai-sdk/anthropic

Configure provider with Helicone

OpenAI Provider
Anthropic Provider
AI Gateway

import { createOpenAI } from "@ai-sdk/openai";

const openai = createOpenAI({
  baseURL: "https://oai.helicone.ai/v1",
  apiKey: process.env.OPENAI_API_KEY,
  headers: {
    "Helicone-Auth": `Bearer ${process.env.HELICONE_API_KEY}`,
  },
});

import { createAnthropic } from "@ai-sdk/anthropic";

const anthropic = createAnthropic({
  baseURL: "https://anthropic.helicone.ai",
  apiKey: process.env.ANTHROPIC_API_KEY,
  headers: {
    "Helicone-Auth": `Bearer ${process.env.HELICONE_API_KEY}`,
  },
});

import { createOpenAI } from "@ai-sdk/openai";

const gateway = createOpenAI({
  baseURL: "https://ai-gateway.helicone.ai",
  apiKey: process.env.HELICONE_API_KEY,
});

Use with AI SDK functions

import { generateText } from "ai";

const { text } = await generateText({
  model: openai("gpt-4o-mini"),
  prompt: "What is the capital of France?",
});

console.log(text);

Streaming Chat

Integrate Helicone with Next.js API routes:

// app/api/chat/route.ts
import { createOpenAI } from "@ai-sdk/openai";
import { streamText } from "ai";

const openai = createOpenAI({
  baseURL: "https://oai.helicone.ai/v1",
  apiKey: process.env.OPENAI_API_KEY!,
  headers: {
    "Helicone-Auth": `Bearer ${process.env.HELICONE_API_KEY}`,
  },
});

export async function POST(req: Request) {
  const { messages } = await req.json();

  const result = await streamText({
    model: openai("gpt-4o-mini"),
    messages,
  });

  return result.toDataStreamResponse();
}

Client-side:

// app/page.tsx
"use client";

import { useChat } from "ai/react";

export default function Chat() {
  const { messages, input, handleInputChange, handleSubmit } = useChat();

  return (
    <div>
      {messages.map((m) => (
        <div key={m.id}>
          {m.role}: {m.content}
        </div>
      ))}

      <form onSubmit={handleSubmit}>
        <input value={input} onChange={handleInputChange} />
        <button type="submit">Send</button>
      </form>
    </div>
  );
}

Session Tracking

Track multi-turn conversations with session IDs:

import { createOpenAI } from "@ai-sdk/openai";
import { generateText } from "ai";
import { v4 as uuidv4 } from "uuid";

const sessionId = uuidv4();

const openai = createOpenAI({
  baseURL: "https://oai.helicone.ai/v1",
  apiKey: process.env.OPENAI_API_KEY!,
  headers: {
    "Helicone-Auth": `Bearer ${process.env.HELICONE_API_KEY}`,
  },
});

const { text } = await generateText({
  model: openai("gpt-4o-mini"),
  prompt: "Hello!",
  headers: {
    "Helicone-Session-Id": sessionId,
    "Helicone-Session-Name": "Customer Chat",
    "Helicone-User-Id": "user-123",
  },
});

In API routes:

// app/api/chat/route.ts
import { createOpenAI } from "@ai-sdk/openai";
import { streamText } from "ai";

const openai = createOpenAI({
  baseURL: "https://oai.helicone.ai/v1",
  apiKey: process.env.OPENAI_API_KEY!,
  headers: {
    "Helicone-Auth": `Bearer ${process.env.HELICONE_API_KEY}`,
  },
});

export async function POST(req: Request) {
  const { messages, sessionId, userId } = await req.json();

  const result = await streamText({
    model: openai("gpt-4o-mini"),
    messages,
    headers: {
      "Helicone-Session-Id": sessionId,
      "Helicone-User-Id": userId,
      "Helicone-Property-Path": req.url,
    },
  });

  return result.toDataStreamResponse();
}

Learn more about Session Tracking.

Custom Properties

Add custom metadata to track important context:

import { generateText } from "ai";

const { text } = await generateText({
  model: openai("gpt-4o-mini"),
  prompt: "Analyze this user query",
  headers: {
    "Helicone-User-Id": "user-123",
    "Helicone-Property-Environment": "production",
    "Helicone-Property-Feature": "query-analysis",
    "Helicone-Property-Version": "v2.1.0",
  },
});

Tool Calling

Helicone tracks tool calls with full visibility:

import { generateText, tool } from "ai";
import { z } from "zod";

const { text, toolCalls } = await generateText({
  model: openai("gpt-4o-mini"),
  prompt: "What's the weather in Paris?",
  tools: {
    weather: tool({
      description: "Get the weather for a location",
      parameters: z.object({
        location: z.string().describe("The city name"),
      }),
      execute: async ({ location }) => {
        // Fetch weather data
        return { temperature: 20, condition: "sunny" };
      },
    }),
  },
  headers: {
    "Helicone-Property-Tools": "weather",
  },
});

// Tool calls are logged with inputs and outputs
console.log(toolCalls);

AWS Bedrock

Use Vercel AI SDK with AWS Bedrock through Helicone:

import { createAmazonBedrock } from "@ai-sdk/amazon-bedrock";
import { generateText } from "ai";

const bedrock = createAmazonBedrock({
  region: process.env.AWS_REGION!,
  accessKeyId: process.env.AWS_ACCESS_KEY_ID!,
  secretAccessKey: process.env.AWS_SECRET_ACCESS_KEY!,
  baseURL: `https://bedrock.helicone.ai/v1/${process.env.AWS_REGION}`,
  headers: {
    "Helicone-Auth": `Bearer ${process.env.HELICONE_API_KEY}`,
    "aws-access-key": process.env.AWS_ACCESS_KEY_ID!,
    "aws-secret-key": process.env.AWS_SECRET_ACCESS_KEY!,
    "Accept-Encoding": "identity",
  },
});

const { text } = await generateText({
  model: bedrock("anthropic.claude-3-5-sonnet-20240620-v1:0"),
  prompt: "Hello!",
  headers: {
    "Helicone-Session-Id": "bedrock-session-123",
  },
});

Response Caching

Reduce costs with response caching:

import { generateText } from "ai";

const { text } = await generateText({
  model: openai("gpt-4o-mini"),
  prompt: "What is 2+2?",
  headers: {
    "Helicone-Cache-Enabled": "true",
  },
});

// Subsequent identical requests return cached responses

Learn more about Response Caching.

Object Generation

Track structured object generation:

import { generateObject } from "ai";
import { z } from "zod";

const { object } = await generateObject({
  model: openai("gpt-4o-mini"),
  schema: z.object({
    name: z.string(),
    age: z.number(),
    city: z.string(),
  }),
  prompt: "Generate a person's profile",
  headers: {
    "Helicone-Property-Type": "structured-generation",
  },
});

console.log(object);

Streaming Objects

Stream structured data generation:

import { streamObject } from "ai";
import { z } from "zod";

const { partialObjectStream } = await streamObject({
  model: openai("gpt-4o-mini"),
  schema: z.object({
    recipe: z.object({
      name: z.string(),
      ingredients: z.array(z.string()),
      steps: z.array(z.string()),
    }),
  }),
  prompt: "Generate a recipe for chocolate cake",
});

for await (const partialObject of partialObjectStream) {
  console.log(partialObject);
}

Use vision models with Helicone:

import { generateText } from "ai";

const { text } = await generateText({
  model: openai("gpt-4o"),
  messages: [
    {
      role: "user",
      content: [
        { type: "text", text: "What's in this image?" },
        {
          type: "image",
          image: "https://example.com/image.jpg",
        },
      ],
    },
  ],
  headers: {
    "Helicone-Property-Type": "vision",
  },
});

Rate Limiting

Control usage per user:

import { generateText } from "ai";

const { text } = await generateText({
  model: openai("gpt-4o-mini"),
  prompt: "Hello!",
  headers: {
    "Helicone-RateLimit-Policy": "100;w=60;s=user",
    "Helicone-User-Id": "user-123",
  },
});

Learn more about Rate Limiting.

Best Practices

Always include session IDs

Track conversations with session IDs:

headers: {
  "Helicone-Session-Id": sessionId,
  "Helicone-User-Id": userId,
}

Use custom properties for filtering

Add context for better organization:

headers: {
  "Helicone-Property-Environment": "production",
  "Helicone-Property-Feature": "chat",
}

Track tool usage

Label requests that use tools:

headers: {
  "Helicone-Property-Tools": "weather,search",
}

Enable caching for repeated queries

Reduce costs with caching:

headers: {
  "Helicone-Cache-Enabled": "true",
}

Server Actions

Use Helicone with Next.js Server Actions:

// app/actions.ts
"use server";

import { createOpenAI } from "@ai-sdk/openai";
import { generateText } from "ai";

const openai = createOpenAI({
  baseURL: "https://oai.helicone.ai/v1",
  apiKey: process.env.OPENAI_API_KEY!,
  headers: {
    "Helicone-Auth": `Bearer ${process.env.HELICONE_API_KEY}`,
  },
});

export async function generateResponse(prompt: string, userId: string) {
  const { text } = await generateText({
    model: openai("gpt-4o-mini"),
    prompt,
    headers: {
      "Helicone-User-Id": userId,
      "Helicone-Property-Source": "server-action",
    },
  });

  return text;
}

Troubleshooting

Requests not appearing in dashboard

Verify your Helicone API key is correct
Check the Helicone-Auth header format: Bearer sk-helicone-...
Ensure you’re passing headers to the provider config, not the individual function
Check browser network tab for failed requests

Streaming not working

Ensure your Next.js API route returns result.toDataStreamResponse()
Check that streaming is enabled in your deployment platform
Verify no middleware is buffering the response

Session tracking not working

Session IDs must be consistent across requests
Pass session ID from client to API route
Use UUIDs or unique strings for session IDs

Headers not being sent

Headers should be in the provider configuration:

// Correct
const openai = createOpenAI({
  headers: { "Helicone-Auth": "..." }
});

// Also supported (per-request)
await generateText({
  model: openai("gpt-4o-mini"),
  headers: { "Helicone-Session-Id": "..." }
});

Next Steps

AI Gateway

Use multiple providers with automatic fallbacks

Session Tracking

Track multi-turn conversations

Custom Properties

Add custom metadata

Response Caching

Reduce costs with caching

Get Started

AI Gateway

Observability

Prompt Management

Evaluation & Testing

Features

Self-Hosting

Integrations

Vercel AI SDK Integration

Integration Methods

Provider Configuration

AI Gateway

Custom Headers

Async Logging

Quick Start

Streaming Chat

Session Tracking

Custom Properties

Tool Calling

AWS Bedrock

Response Caching

Object Generation

Streaming Objects

Rate Limiting

Best Practices

Server Actions

Troubleshooting

Next Steps

AI Gateway

Session Tracking

Custom Properties

Response Caching

Get Started

AI Gateway

Observability

Prompt Management

Evaluation & Testing

Features

Self-Hosting

Integrations

Documentation Index

​Integration Methods

Provider Configuration

AI Gateway

Custom Headers

Async Logging

​Quick Start

​Streaming Chat

​Session Tracking

​Custom Properties

​Tool Calling

​AWS Bedrock

​Response Caching

​Object Generation

​Streaming Objects

​Multi-Modal (Vision)

​Rate Limiting

​Best Practices

​Server Actions

​Troubleshooting

​Next Steps

AI Gateway

Session Tracking

Custom Properties

Response Caching

Integration Methods

Quick Start

Streaming Chat

Session Tracking

Custom Properties

Tool Calling

AWS Bedrock

Response Caching

Object Generation

Streaming Objects

Multi-Modal (Vision)

Rate Limiting

Best Practices

Server Actions

Troubleshooting

Next Steps