◈ Pranab Sona AI
Any provider · Any API key · Every latest model

Chat with every AI model
in one clean place.

Pranab Sona AI is a fast, private chat interface. Bring your own key from OpenRouter, TokenRouter, OpenAI, Groq, DeepSeek, Mistral, Gemini, xAI or any OpenAI-compatible endpoint and instantly talk to GPT, Claude, Gemini, Llama, Grok, DeepSeek and more — with streaming, vision, voice, prompt templates and model comparison. Nothing is stored on a server.

Try the demo
🔒 Your key and chats stay in your browser's local storage.

Try a free model right here

An open-source AI model runs entirely in your browser — no API key, no login, fully private. The first reply downloads the model, then it's cached.

🔒 Runs 100% in your browser via WebLLM — nothing is sent to any server.

Everything you need, nothing you don't

A focused chat experience built for people who want the latest models without the clutter.

🌐

Any provider, any key

OpenRouter, OpenAI, Groq, DeepSeek, Mistral, Gemini, xAI or a custom endpoint. Live model list, updated automatically.

⚡

Streaming replies

Token-by-token so you read the answer as it's written, with a stop button.

🔀

Compare models

Run the same prompt against two models side by side.

🖼️

Vision & files

Attach images for vision models, or PDFs/text files for context.

🎙️

Voice in & out

Dictate with your mic and have replies read aloud.

📚

Prompt library

Save and reuse your favourite prompts in one click.

🎨

Themes

Dark or light, custom accent colour, and adjustable font size.

📤

Export & share

Download chats as Markdown/JSON, print to PDF, or share a link.

🔐

Private by design

Your key and conversations never leave your browser.

Hundreds of models, one interface

A few of the families you can reach across supported providers.

OpenAI GPTAnthropic Claude Google GeminiMeta Llama MistralDeepSeek xAI GrokQwen CohereNous Hermes Perplexity+ hundreds more

FAQ

Is it free?

The app is free. You pay your chosen provider directly for model usage with your own key. Some providers (like OpenRouter) offer free models too.

Which providers work?

Any OpenAI-compatible API: OpenRouter, TokenRouter, OpenAI, Groq, DeepSeek, Mistral, Together, Fireworks, xAI (Grok), Perplexity and Google Gemini. Pick Custom in Settings to point at any other base URL, including a local server.

Where do I get an API key?

From your provider's dashboard — for example openrouter.ai/keys, OpenAI, Groq or Google AI Studio. Settings links straight to the right page for each provider.

Where is my API key stored?

Only in your browser's localStorage. It's sent only to the provider you select, never to us.

Do my chats sync across devices?

Not currently — everything is stored locally. You can export your data and import it on another device.

Which models support images?

Any vision-capable model your provider offers (e.g. GPT-4o, Claude, Gemini). Attach an image and pick one of those.