free4chat

free4.chat — real-time voice + text chat. No sign-up, no server to run. Open a room and talk.

⚠️ Personal project / experimental. Use at your own risk.

Features

• 🎙️ Voice chat in rooms
• 💬 Text chat with emoji
• 📎 File & image transfer (inline preview)
• 🖥️ Screen sharing
• 🤖 Luna — optional AI assistant (mention @luna to invoke)
• 🔒 No accounts, no persistent data
• 🛡️ Cloudflare Turnstile bot protection

Privacy

free4chat is built around two principles: no data outlives the conversation, and you don't need to trust any server.

What we don't store:

• No accounts, no sign-up, no identity
• Messages exist only in participants' browser memory — close the tab and they're gone
• Files and images are transferred via WebRTC data channels, never written to any database
• Voice is relayed through Cloudflare's media nodes but never recorded

What does persist (and why it's fine):

• A room name → meeting ID mapping is kept in Cloudflare KV with a 30-day TTL so rejoining the same room name works. It contains no messages, no users, no content.
• When Luna AI is enabled, messages sent to @luna are transmitted to an external AI model (Cloudflare AI Gateway → @cf/zai-org/glm-4.7-flash) for processing. The last 20 messages of conversation context are retained in a Durable Object for the lifetime of the room session only. Luna is opt-in and disabled by default.
• Your nickname is saved in browser localStorage for convenience. Clear it anytime.

The Worker's only job is to issue a short-lived auth token — after that, all communication is peer-to-peer or via Cloudflare's media plane with no application-layer logging.

Tech Stack

Layer	Technology
Frontend	Next.js 15, Tailwind CSS, Cloudflare RealtimeKit React SDK
API	Next.js API routes deployed as Cloudflare Worker via @opennextjs/cloudflare
AI	BotSession Durable Object → Cloudflare AI Gateway → @cf/zai-org/glm-4.7-flash
Storage	Cloudflare KV (room metadata, rate limiting) + DO KV storage (Luna chat history)
Media	Cloudflare RealtimeKit (WebRTC, audio, data channels, screen sharing)
Security	Cloudflare Turnstile (full-page bot challenge) + origin whitelist + KV rate limiting

Stack History

This project has gone through three stacks, always with the same product goal:

Branch	Stack	Why it changed
golang	Go + Pion WebRTC + coturn	Self-hosted infra is too much overhead for a personal project
elixir	Elixir + Membrane Framework	Maintaining a server cluster is still heavy for something this small
cloudflare (this branch)	Cloudflare RealtimeKit + Workers	Fully serverless — no servers to manage, file transfer built-in, free tier works

The product never changed. The ops burden did.

The full story — WebRTC internals, why each stack was chosen, where AI voice bots are headed — is written up here: 一个 WebRTC 聊天室的三次演进 (Chinese)

Development

See DEVELOPMENT.md for local setup, deployment, and architecture notes.

License

MIT