Intelligence, made yours.

Think with Zexir. Create with Zevol. Train with Zeneral.
Run locally with Zera.

The Zexiry mark, suspended above a softly lit ivory horizon

A MORE PERSONAL INTELLIGENCE

Built for the way
you think.

Zexiry brings together a family of focused AI tools designed to help you think deeper, create freely, learn continuously, and run on your terms.

ZEXIR · THINK

One place. More possibilities.

A conversation can become a clearer idea, a deeper answer, or your next step.

Meet Zexir

ZEVOL · CREATE

An idea is a beginning.

Describe what you want.
Preview it, refine it, and publish it.

Explore Zevol

ZENERAL · TRAIN

A lighter path to training.

An AI architecture designed to make model training work with less VRAM.

Inside Zeneral

ZERA · LOCAL

Your AI. On your Mac.

A private 24B model that runs locally.
A different way to make AI yours.

Download Zera

Zexir Auto

Help me understand this idea.

Let’s make it clearer.

Start with what matters. Break it into smaller parts. Then choose a useful next step.

The idea, in simple terms
The questions worth asking
A practical way forward

Ask Zexir

An illustrative look inside Zexir

ZevolPreview

Atelier Objects · About

HANDMADE, WITH INTENTION

Objects for
a calmer home.

Everyday pieces.
Made to be kept.

Explore the collection

Your idea

A calm website for my ceramics studio.

Your first preview is ready. What would you like to refine?

Make it yours

An example of what you can create

Three layers of glass, ivory, and graphite, representing Zeneral's architecture

A silver laptop on a warm ivory surface, representing local AI on your Mac

ZEXIR

Make space
for better ideas.

Zexir brings conversation, research, and practical tools into one workspace. Ask a quick question, explore a difficult topic, work through a document, or turn an answer into something you can use.

Open Zexir

Start with a question.
Keep going with possibilities.

Choose a model, let Auto route your conversation, or use your saved Model Studio preferences. The tools stay close to the conversation, so you can move between thinking, searching, and creating without starting over.

More than one perspective

Multi-Model Research brings selected models into the same request and combines their contributions. Choose the models you want to use, or enable your selection automatically for future requests. Availability depends on your plan.

Research that reaches further

Search for current information and follow sources from the answer. Use research when a question needs evidence, context, or a view beyond a model’s stored knowledge.

Your model preferences

Model Studio lets you save how your requests are routed. Use the model chooser to return to those preferences, or select a model for the conversation in front of you.

Work in more ways

Type, dictate, use voice chat, or bring files into a conversation. Keep useful work in your Library and connect available plugins, including X, through the Plugins page.

Make code tangible

Run supported HTML and Python snippets from the answer. HTML previews appear in place of the code, with controls to pause, continue, or stop the run.

A workspace that remembers

Saved chats and preferences follow your account. Drafts stay with the correct chat. For a conversation you don’t want in your saved history, switch to Private mode.

Web and iOS. Free to start, with additional models and capabilities on paid plans. Explore plans

ZEVOL

Give your idea
somewhere to live.

You know what you want to make. Zevol gives you a place to describe it, see it take shape, and keep refining it through conversation.

Open Zevol

Describe. Preview. Refine.

Start with your idea in plain language. Zevol creates a web app preview alongside the conversation. Try it, ask for changes, and build on what is already there. You can pause a response while you decide what to do next.

A home for every app

Apps act as folders for their own chats. Start a new conversation inside an app to explore another change while keeping the work together.

See what you’re making

The live preview sits beside your request. Move between the conversation and the result, and check how a responsive app looks on supported screen sizes.

Publishing through conversation

When you want to share your app, ask Zevol to deploy it. Publishing uses the configured deployment service; a custom domain follows a separate availability, pricing, and approval flow.

The details stay clear

Review the exact price and purchase terms before authorizing a domain purchase. Receipts stay in Zevol Settings. Domain purchasing depends on the registrar connection being configured.

Designed for self-contained web apps, with a live preview and a conversation that stays with your project.

ZENERAL

Think beyond
the memory limit.

Zeneral is Zexiry’s AI architecture focused on training models with less VRAM. Its direction is simple: make model development more accessible on constrained hardware.

Talk to us about Zeneral

A different starting point
for AI architecture.

GPU memory shapes what people can train and where they can experiment. Zeneral puts that constraint at the center of its design: an architecture intended for lower-VRAM model training.

The goal is to open more room for experimentation without making access to a large GPU setup the first step. This is a training-focused part of the Zexiry family, distinct from Zexir’s chat workspace and Zera’s local inference.

Zeneral is an ongoing architecture effort. For technical details, supported workloads, and hardware requirements, get in touch with our team.

ZERA

Intelligence
closer to home.

A private 24B model that runs directly on your Mac. Download the model and use a compatible local runtime to bring AI onto your own device.

Download Zera

A laptop representing Zera's local runtime

Local is a different kind of freedom.

Zera is available as a 24B, 4-bit MLX model download. Running the model locally means inference happens in your local runtime rather than through a hosted chat model.

Use Zera with a compatible MLX runtime on your Mac. Allow enough memory and storage for the complete model, then choose a local setup that fits the way you work.

View the model on Hugging Face

WHY ZEXIRY

More room
for what comes next.

We’re building a family of tools around one idea: intelligence should help people do more with what they already have.

Zexir makes a conversation useful. Zevol gives an idea a place to become real. Zeneral explores a lighter path to training. Zera brings a model closer to home.

Different starting points. A shared direction.

mail@zexiry.com

Start with
what’s on your mind.

Try Zexir