Browser Edition

The browser edition runs the full Hadrian Gateway in your browser using WebAssembly. No installation, no server, no Docker — just open the page and start chatting.

Quick Start

1. Open app.hadriangateway.com
2. Connect a provider (OpenRouter is one click, or add your own API keys)
3. Start chatting

Using Ollama

Ollama lets you run LLMs locally for free. Because the browser edition makes API calls directly from your browser, Ollama needs to be configured to allow requests from the Hadrian origin.

1. Install Ollama

Download and install from ollama.com.

2. Enable CORS

By default, Ollama only allows requests from localhost. The browser edition runs at a different origin, so you need to set the OLLAMA_ORIGINS environment variable.

macOS

If you run Ollama as an application:

launchctl setenv OLLAMA_ORIGINS "https://app.hadriangateway.com"

Then restart the Ollama application. To allow all origins (less secure, but convenient for development):

launchctl setenv OLLAMA_ORIGINS "*"

Linux (systemd)

sudo systemctl edit ollama.service

Add the following under the [Service] section:

[Service]
Environment="OLLAMA_ORIGINS=https://app.hadriangateway.com"

Then reload and restart:

sudo systemctl daemon-reload
sudo systemctl restart ollama

Linux / macOS (manual)

If you run ollama serve directly in a terminal:

OLLAMA_ORIGINS=https://app.hadriangateway.com ollama serve

Windows

1. Quit Ollama from the taskbar
2. Open Settings → search for "Edit environment variables for your account"
3. Add a new variable: OLLAMA_ORIGINS = https://app.hadriangateway.com
4. Click OK and restart Ollama

Docker

docker run -e OLLAMA_ORIGINS=https://app.hadriangateway.com -p 11434:11434 ollama/ollama

3. Pull Models

Ollama only serves models that have been downloaded. Pull some models to get started:

# Small and fast — great for quick tasks
ollama pull llama3.2

# Larger and more capable
ollama pull llama3.3

# Reasoning model
ollama pull deepseek-r1

# Google's open model
ollama pull gemma3

# Lightweight and efficient
ollama pull phi4

# Multilingual with tool use
ollama pull qwen3

See the full list of available models at ollama.com/library.

4. Connect in Hadrian

Open the setup wizard (or re-open it from the user menu). If Ollama is running and CORS is configured, the wizard will auto-detect it and show a Connect button.

If Ollama is not detected, check that:

• Ollama is running (ollama serve or the Ollama app)
• OLLAMA_ORIGINS includes the Hadrian origin
• Ollama is accessible at http://localhost:11434

Limitations

The browser edition has some limitations compared to the server version:

• No real-time streaming — responses are buffered (the full response appears at once)
• No usage tracking — token usage is not recorded
• No caching, rate limiting, or budgets
• Requires a modern browser — Chrome 91+, Edge 91+ (module service workers)
• Provider API keys are stored in your browser — they never leave your device, but are only as secure as your browser's storage

For teams, SSO, guardrails, and full feature support, use the server version.