New Chat

The Universal AI Chat

Chat with your favorite LLMs. ChatGPT, Claude, Google Gemini, Grok... All in one place.
Bring Your Own Key (BYOK) and enjoy any AI model in a single interface.

Local Database

History is stored locally in your browser. Only you have access.

BYO Keys

Pay providers directly. No middleman markup.

Backup & Sync

Sync seamlessly across devices via Cloud Storage. Never lose your prompts.

Tree & Branching

Organize chats in nested folders. Branch conversations to explore different outcomes.

Optimized Uploads

Files are pre-processed to save tokens and improve LLM comprehension.

Prompt Library

Save your best prompts. Reuse them with variables and shortcuts.

Snappy UI

No bloat, just speed. What you need, where you want it. Rich features with minimal noise.

Deep Customization

Tokyo Night, Catppuccin, Light/Dark modes, font scaling, and fluid layout options.

Rich Formatting

Beautiful syntax highlighting, code rendering, and markdown tables.

How it works

1
Enter your API Key

Click settings or paste your key when prompted. It is encrypted in LocalStorage.

2
Select a Model

Switch between GPT-5.2, Claude 4.5, Gemini 3, or any other AI model instantly.

3
Start Chatting

Type below to begin your first conversation.

Frequently Asked Questions

Is ChatStack AI free to use?

Yes, the interface is free and open source. You only pay the AI providers (OpenAI, Anthropic, Google, xAI, etc.) directly for the API usage. This is typically much cheaper than monthly subscriptions for casual to moderate users.

Which AI providers and models are supported?

ChatStack AI supports the latest flagship models from all major providers. You can seamlessly switch between:

  • OpenAI: GPT-5.2, GPT-5.2 Pro, and Codex Max
  • Anthropic: Claude 4.5 Sonnet & Opus
  • Google: Gemini 3.0 Pro & Flash
  • xAI: Grok 4.1 & Vision
  • DeepSeek: DeepSeek V3.2 & Coder
  • Alibaba: Qwen 3 Max, Plus, Flash & Coder
  • Meta: Llama 4
  • Mistral: Mistral 3
  • Local: Any model via Ollama

Check the full list of supported parameters and prices on our Model Specifications page.

How does 'Bring Your Own Key' work?

Instead of paying us a subscription, you generate an API Key from the AI provider's developer dashboard. You paste that key into ChatStack settings. This gives you full access to the models at wholesale prices without middleman markups.

Can I sync my chats across devices?

Yes. While we are local-first, we offer a Cloud Sync feature. It uses your own Document Store credentials to securely synchronize your history, folders, and prompts between your desktop and mobile devices.

Is my chat history private?

Absolutely. ChatStack AI uses a "Local First" architecture. Your chat history, API keys, and prompts are stored in IndexedDB within your browser. Data is never sent to our servers, only directly to the AI provider you select for inference.

What are the risks of local storage?

Since data lives in your browser, clearing your "Site Data" or "Cache" can delete your history. We strictly recommend using our Export Data feature regularly (JSON format) or subscribing to the Cloud Sync service to ensure your prompts and history are backed up safely.

Can I save custom prompts?

Yes. You can build a personal library of reusable prompts. You can define System Prompts (to change how the AI behaves) and regular Snippet Prompts (for repetitive tasks). Variables are supported for dynamic insertion.

How do folders and branching work?

Organize your chats into nested folders and subfolders. Inside a chat, you can "Edit" any previous message to create a Branch. This allows you to explore different outcomes from a single point in the conversation without losing the original context.

Is the interface customizable?

Fully. Choose from multiple themes (Tokyo Night, Catppuccin, Gruvbox), adjust font sizes, and toggle between "Focus Mode" or "Power User Mode". The interface is designed to feel native and snappy.

How are file uploads handled?

We optimize documents before sending them to the LLM. PDF, CSV, DOCX, and Code files are parsed and stripped of unnecessary whitespace to save you tokens (money) and improve the model's ability to focus on the content.