Deploy your ChatGPT SDK integration to Vercel with proper environment configuration

Vercel Deployment Guide

Deploy your ChatGPT SDK application to Vercel with optimized configuration for AI streaming and MCP endpoints.

Pre-Deployment Checklist

OpenAI API key generated
All environment variables documented
Streaming endpoints tested locally
Build passes without errors: bun run build

Environment Variables

Configure these in your Vercel project settings:

Variable	Description	Required
`OPENAI_API_KEY`	Your OpenAI API key	Yes
`OPENAI_MODEL`	Default model (e.g., `gpt-4o`)	No
`MAX_TOKENS`	Maximum tokens per response	No
`ENABLE_STREAMING`	Enable response streaming	No

Production API Key Best Practices

# Use a restricted API key with limited permissions
# Only allow access to required models
# Set IP restrictions if possible
# Rotate keys regularly

Vercel Configuration

vercel.json

Create or update vercel.json for optimal AI streaming:

{
  "functions": {
    "app/api/chat/route.ts": {
      "maxDuration": 60
    }
  },
  "headers": [
    {
      "source": "/api/chat",
      "headers": [
        {
          "key": "Cache-Control",
          "value": "no-store, max-age=0"
        }
      ]
    }
  ]
}

Edge Runtime (Optional)

For lower latency, use Edge Runtime:

// app/api/chat/route.ts
export const runtime = 'edge';
export const maxDuration = 60;

import { openai } from '@ai-sdk/openai';
import { streamText } from 'ai';

export async function POST(req: Request) {
  // Edge-compatible implementation
}

Deployment Steps

Option 1: Vercel CLI

# Install Vercel CLI
bun add -g vercel

# Login
vercel login

# Deploy
vercel --prod

Option 2: Git Integration

Push your code to GitHub/GitLab/Bitbucket
Import project in Vercel dashboard
Configure environment variables
Deploy

Post-Deployment Verification

Test the Chat Endpoint

curl -X POST https://your-app.vercel.app/api/chat \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Monitor Logs

Check Vercel Functions logs for:

API response times
Error rates
Token usage patterns

vercel logs --json

Performance Optimization

Caching Strategies

// Cache tool results when appropriate
const cache = new Map();

const result = streamText({
  model: openai('gpt-4o'),
  messages,
  tools: {
    getUserData: {
      description: 'Get user data',
      parameters: z.object({ userId: z.string() }),
      execute: async ({ userId }) => {
        if (cache.has(userId)) {
          return cache.get(userId);
        }
        const data = await fetchUserData(userId);
        cache.set(userId, data);
        return data;
      },
    },
  },
});

Rate Limiting

import { Ratelimit } from '@upstash/ratelimit';
import { Redis } from '@upstash/redis';

const ratelimit = new Ratelimit({
  redis: Redis.fromEnv(),
  limiter: Ratelimit.slidingWindow(10, '1 m'),
});

export async function POST(req: Request) {
  const ip = req.headers.get('x-forwarded-for');
  const { success } = await ratelimit.limit(ip ?? 'anonymous');
  
  if (!success) {
    return new Response('Rate limit exceeded', { status: 429 });
  }
  
  // Continue with chat logic
}

Troubleshooting Deployment Issues

Issue	Solution
504 Timeout	Increase `maxDuration` in vercel.json
Streaming not working	Ensure `runtime` is set to `edge` or `nodejs`
CORS errors	Add CORS headers to API routes
High latency	Consider Edge Runtime or regional deployment

Next Steps

Troubleshooting Guide
Monitor your deployment analytics
Set up alerting for errors

Vercel Deployment

On this page