🌐 Languages: English | فارسی
One Chat is a lightweight AI chat interface designed to work with both local models and cloud AI services. Instead of using a different interface for every provider, you can open a single page, connect your preferred model, and start chatting.
The entire application is contained in a single HTML file. There is nothing to install, no build process, and no backend server required.
One Chat supports the following providers out of the box:
- Ollama
- LM Studio
- OpenRouter
- OpenAI
- Groq
- Together AI
- Fireworks AI
- DeepSeek
- Google Gemini
- Chrome AI (Gemini Nano)
- 9Router
- Any OpenAI-compatible API
Local providers such as Ollama and LM Studio usually do not require an API key. Cloud providers require the corresponding API key.
- Lightweight single-file application with no dependencies
- Connect to both local and cloud AI models
- Streaming responses in real time
- Markdown rendering with tables, lists, blockquotes, and code blocks
- Copy buttons for responses and code snippets
- Compare mode for querying two models side by side
- Configurable System Prompt, Temperature, and Max Tokens
- Upload text files, JSON, Markdown, and images
- Save and search chat history
- Export conversations as Markdown
- Backup and restore application data
- Built-in prompt library
- English and Persian interface
- Light and dark themes
- Responsive design for desktop and mobile
- Download
one-chat.html. - Open it in a modern browser such as Chrome, Edge, or Firefox.
- Open Settings and select your provider.
- Enter the API endpoint, model name, and API key (if required).
- Click Test Connection, then start chatting.
That's it—no installation or build step is required.
Make sure Ollama is running and the model you want to use is already available.
In One Chat:
- Select
Ollamaas the provider. - Leave the default endpoint as
http://localhost:11434/v1. - Enter your installed model name (for example,
llama3.2). - Click Test Connection.
If your browser blocks the connection, enable CORS for the page origin in your Ollama configuration.
Start the local server in LM Studio, then:
- Select
LM Studioas the provider. - Verify the default endpoint:
http://localhost:1234/v1. - Enter the loaded model name.
- Click Test Connection.
If the connection fails, check the LM Studio server's CORS settings.
If your provider exposes an OpenAI-compatible API, select OpenAI Compatible and enter:
- The API base URL (usually ending with
/v1) - The exact model name
- Your API key (if required)
You can also use the Additional JSON Parameters field to pass options such as top_p.
Chats and settings are stored locally in your browser. One Chat does not use an intermediary server—requests are sent directly from your browser to the selected provider.
If Save API Key is disabled, your API key is kept only until the browser tab is closed. If enabled, the key is stored in your browser's local storage. For security reasons, avoid enabling this option on shared or public computers.
When creating a backup, you can choose whether API keys should be included. For better security, it is recommended to exclude them.
If you cannot connect, check the following:
- Make sure the local model or server is running.
- Verify the API endpoint and model name.
- Ensure your API key is valid.
- Confirm that your browser can access the model endpoint.
- Enable CORS on your local server if necessary.
- If you opened the application using
file://and your provider blocks the connection, serve the file through a simple local web server.
One Chat is not intended to replace large, feature-rich AI clients. It is designed to be a simple, lightweight, and independent interface that lets you connect to your preferred AI model and start chatting in seconds.