Next.js Gemini Chatbot

The Next.js Gemini Chatbot is an open-source AI chatbot template from Vercel Labs that combines the Vercel AI SDK with Google Gemini models on a Next.js App Router foundation. It takes a Google Generative AI API key and optional authentication secrets as input and produces a deployable chat application with streaming responses, generative UI, and persistent chat history. The template runs on Next.js 14 and can be deployed to Vercel with a single click or run locally with pnpm dev.

What is the Next.js Gemini Chatbot?

The Next.js Gemini Chatbot is a GitHub template built by Vercel Labs that scaffolds a production-ready AI chat application using the Vercel AI SDK and Google Gemini gemini-1.5-pro as the default model. It accepts an API key and environment variables, then outputs a full chat interface with React Server Components, Server Actions, and NextAuth authentication. The template is officially maintained by Vercel Labs in the vercel-labs/gemini-chatbot repository, which has 1,347 stars on GitHub.

Key Features

• Next.js App Router — Uses advanced routing, React Server Components, and Server Actions for server-side rendering and increased performance.
• AI SDK integration — Unified API for generating text, structured objects, and tool calls with LLMs; supports Google (default), OpenAI, Anthropic, Cohere, and many more providers.
• shadcn/ui styling — Built on Tailwind CSS with component primitives from Radix UI for accessibility and flexibility.
• Data persistence — Vercel Postgres powered by Neon saves chat history and user data; Vercel Blob provides efficient object storage.
• NextAuth.js authentication — Simple and secure authentication setup.
• One-click Vercel deploy — Deploy button clones the repo with required environment variables (AUTH_SECRET, GOOGLE_GENERATIVE_AI_API_KEY) and creates Postgres and Blob stores automatically.

Who is it for?

• Developers building AI chatbots — Use the template to skip configuration and go straight to customizing chat logic, streaming, and generative UI.
• Next.js teams — Learn how the AI SDK integrates with React Server Components, Server Actions, and route handlers.
• Startups prototyping on Vercel — Deploy a working chatbot in minutes, attach authentication, and persist conversations with Vercel Postgres.

What can you do with the Next.js Gemini Chatbot?

• Deploy your own AI assistant — Clone the repo to Vercel, add your Google API key, and get a live chatbot with a shareable URL.
• Switch model providers — Change from Gemini to OpenAI, Anthropic, or Cohere with just a few lines of code via the AI SDK.
• Build generative UI — Use the AI SDK's hooks to render tool calls, structured objects, and dynamic interfaces inside chat messages.
• Run it locally — Install the Vercel CLI, pull environment variables with vercel env pull, then start with pnpm install and pnpm dev on localhost:3000.

How does the template work?

Deployment is one-click via Vercel's clone flow, which prompts for the AUTH_SECRET and GOOGLE_GENERATIVE_AI_API_KEY environment variables and provisions a Postgres database and Blob storage. For local development, the README instructs you to link your project with vercel link, download secrets with vercel env pull, then run pnpm install and pnpm dev. The app then connects to Gemini through the AI SDK, stores conversations in Postgres, and uses NextAuth for authentication.

FAQ

Do I need a Google API key to use this template?

Yes, the template uses Google Gemini models by default, so you need a GOOGLE_GENERATIVE_AI_API_KEY. You can also switch to OpenAI, Anthropic, or Cohere by changing the provider in the AI SDK configuration.

Can I deploy without a database?

The template is designed to use Vercel Postgres for chat history. If you omit the database, the chat will still function but history persistence and user data storage will not work.

Is the template free?

The template itself is open-source and free, but you pay for any API usage (Google Gemini, OpenAI, etc.) and for Vercel storage and serverless functions if you exceed free plan limits.