Browser-native · Zero backend · Free forever

Chat with every AI model at once.

Duplex is a free, browser-only frontend that connects every AI endpoint you own — local or cloud — and streams all their responses side by side. No login. No server. No data leaving your device.

8 providers
out of the box
Parallel streams
all at once
100% local
no server needed
Open source
MIT license

What is Duplex?

Duplex is a universal AI frontend that lives entirely in your browser. Connect your local Ollama server, your OpenAI key, your Anthropic key — as many sources as you want.

Every response streams back in its own panel, side by side. Compare tone, accuracy, speed, and reasoning across models in real time. No backend, no account, no data leaving your device.

Browser-native

Runs entirely in your browser. No server, no backend.

Zero setup

Add your keys or point to your local server. Go.

Privacy first

Your keys and prompts never leave your device.

Open source

MIT licensed. Audit the code yourself.

Features

Everything you need. Nothing you don't.

Simultaneous streaming

One prompt, all endpoints, all at once. Every response streams token by token in its own panel — no waiting for one to finish before the next starts.

8 providers built in

OpenAI, Anthropic, Groq, Gemini, xAI Grok, Perplexity, OpenRouter, and any local OpenAI-compatible server. Add keys, toggle on, go.

Local model support

Point Duplex at your Ollama, LM Studio, or llama.cpp server. Configure host and port. Run completely offline if you want.

Full conversation history

Every conversation is persisted in localStorage. Titles auto-generate from your first message. Switch between chats instantly.

Zero server footprint

Duplex is a static site. No analytics, no telemetry, no backend. Your prompts and keys exist only between your browser and your chosen AI providers.

Open source, MIT

Fork it, self-host it, modify it. The entire codebase is public. Audit the key storage logic yourself — it takes 30 seconds.

How it works

Up and running in four steps.

1

Add your endpoints

OpenAI, Anthropic, Groq, Gemini, xAI, Perplexity, OpenRouter, or local Ollama.

2

Toggle them on

Activate the endpoints you want to compare.

3

Send a prompt

Type your message and hit enter.

4

Compare in real time

Watch every model respond simultaneously, side by side.

Providers

Works with every major provider.

OpenAIsk-...
Anthropicsk-ant-...
Groqgsk_...
GeminiAIza...
xAIxai-...
Perplexitypplx-...
OpenRoutersk-or-...

Your data stays exactly where you put it.

Duplex is a static site. No backend. No database. No analytics. No telemetry. Your API keys are stored only in your browser, obfuscated. Your prompts never leave your device.

Zero backend

Duplex is a static site. No server to compromise.

Obfuscated keys

Your API keys are encrypted in localStorage.

No tracking

Zero analytics, zero telemetry, zero tracking.

Open source

Audit the code yourself. MIT licensed.

Before you launch

Quick pre-launch checklist.

A modern browser (Chrome, Firefox, Safari, Edge) — no install required

At least one API key OR a running local model server (Ollama, LM Studio)

If using Ollama: CORS enabled (one command — see the docs)

Know which model you want to use — or use the defaults

Internet connection for cloud providers; offline works for local-only

FAQ

Common questions.