Private AI. Built from the ground up.
An AI assistant designed for users who need broad model access without prompt logging, data training, or unnecessary content restrictions.
Privacy by default
Every conversation on Meriq operates under a zero-retention policy. Your prompts and responses are not logged, stored for training, or shared with third parties. This is not a toggle you have to find in settings. It is the default behavior for every user on every plan.
For users who want persistent conversations, Meriq offers an optional cloud vault that encrypts all stored data with AES-256-GCM encryption. The encryption key is derived from your credentials and never stored on our servers. If you don't enable the vault, nothing persists after your session ends.
Broad model access
Meriq gives you access to a wide range of AI models. Switch models mid-conversation, and if one becomes momentarily unavailable, your request continues without interruption.
Chat
Streaming chat with a range of fast and reasoning models, and seamless mid-conversation switching.
Deep Research
Autonomous multi-step web research with cited reports.
Agent Mode
Code execution, web browsing, and file operations inside an isolated workspace.
Image Generation
Image generation with built-in prompt refinement.
File Analysis
Upload PDFs, images, and code for full-context analysis.
What Meriq does differently
| Feature | Typical chatbot | Meriq |
|---|---|---|
| Prompt retention | Stored on servers | None |
| Training on user data | Usually opt-out, by default in | Never |
| Conversation encryption | Server-side encryption | AES-256-GCM end-to-end vault |
| Model switching | Single model per session | Mid-conversation, instant |
| Reliability | Single point of failure | Automatic continuity — your chat stays up even when one model has a hiccup |
| Content policy | Heavy filtering | Broad access for lawful use |
| Refusal handling | Quota burned, no recourse | Automatic refund and retry |
| Free tier | Often trial or none | Yes, no credit card required |