World's First AI-Made AI App · Completely Free

The AI chat app built by AI.

Local models. Voice cloning. Coding agent. Video generation. No subscriptions.

↓ Download Aether See all features →

108Local Models

100%Free Forever

0Cloud Required

∞Messages

Aether — New conversation

New conversation

● Ready · Aether Code

✦

What can I help with?

Ask me anything — I think, write, code, analyse, and explore ideas with you.

🌸 Explain a concept

Neural networks, quantum physics...

💻 Write code

Debug, refactor, or build something

💡 Brainstorm ideas

Projects, names, strategies...

📚 Summarise & research

Books, topics, documents...

Message Aether...

→

The world's first AI chat app designed, built, and shipped entirely by an AI. Every feature. Every line of code. Every pixel.

And it actually works. Remarkably well.

💬 Local AI Chat

💻 Codex Coding Agent

🎤 Live Voice Assistant

🖼️ Image Generation

🎬 Video Generation

🗣️ Voice Cloning

🎵 Music Generation

🧠 108 Local Models

🌐 Web Search

☁️ Cloud Models

🔆 4× Upscaling

🤖 Autonomous Agents

💬 Local AI Chat

💻 Codex Coding Agent

🎤 Live Voice Assistant

🖼️ Image Generation

🎬 Video Generation

🗣️ Voice Cloning

🎵 Music Generation

🧠 108 Local Models

🌐 Web Search

☁️ Cloud Models

🔆 4× Upscaling

🤖 Autonomous Agents

Everything included

Not just a chat window.

A full local AI studio — coding agent, voice assistant, image generation, video generation, voice cloning, and more. All free.

💻

Codex — Coding Agent

Open any folder. Ask it to build something. Codex reads, writes, and creates files autonomously. Supports timed work sessions — say "you've got 45 minutes" and watch it go.

›generate a full snake game with CSS

✓ Written: games/snake.js

✓ Written: games/index.html

✓ Created dir: css/

✓ Written: css/game.css

›

✦ Aether Codex

🎤

Live Voice Assistant

Hands-free AI with wake word detection. Say "Hey Aether" and start talking. Replies are spoken back in a cloned voice. Enable in Settings → System.

Wake wordFully offlineVoice cloning

🧠

108 Local Models

108

Qwen3, DeepSeek R1, Gemma3, Phi-4, Llama3.3 and more — all running locally.

🖼️

Image Generation

Say "generate an image of..." in any chat. ComfyUI powers FLUX.1 locally. 1024×1024 default.

FLUX.11024×1024Local

🎬

Video Generation

Generate short video clips locally via ComfyUI workflows. Multiple style presets. Seed locking.

CinematicAnimeRealistic

🗣️

Voice Cloning & TTS

Clone any voice from a short sample using XTTS. Or use Piper for fast offline TTS. Aether reads its own replies back to you in the cloned voice — no cloud required.

XTTS ClonePiper TTS100% OfflineMulti-language

🤖

Custom AI Profiles

Upload your own documents, set a persona, pick a base model. Aether builds a specialist AI trained on your files — your private expert, running locally.

PDF / TXT / MDSpecialistFrom Scratch

☁️

Cloud Models

Add your OpenAI, Anthropic, or OpenRouter key. Switch between local and cloud in one click.

GPT-4.1ClaudeOpenRouter

🌐

Web Search

Live Brave Search integration. Ground answers in real results. 1,000 searches/month free.

Brave APIAuto-search

🔆

Image Upscaling

4× upscale with RealESRGAN via ComfyUI. Restore detail in any image completely locally.

RealESRGAN4× boostOffline

🎵

Music & Sound Effects

Generate music and sound effects locally using ComfyUI audio nodes. Describe what you want — Aether produces the audio file directly. No cloud, no subscription.

Music genSound FXComfyUI

🎭

Personalities & Agents

Give the AI a persistent persona pinned to a specific model. Or launch autonomous Agents that plan, execute, and iterate across entire projects — with human-approval gates so you stay in control.

PersonalitiesAutonomous agentsApproval gates

Zero friction setup

Up and running in minutes.

No Docker. No Python venv. No terminal commands. Download, install, done.

⬇️

Download & Install

Grab the installer for your OS. Double-click and follow the wizard. Aether is a single native app — no dependencies to manage, no command line required.

~5 min · Windows / macOS / Linux

🧠

Pick Your First Model

Open the Models tab. Aether detects your GPU and recommends the best model for your hardware. One click to download. It'll be ready in minutes.

Qwen3, Llama, DeepSeek & 105 more

✦

Start Chatting

That's it. Open a conversation, type your first message, and watch your local AI respond — privately, instantly, completely free. No account. No telemetry. No limits.

100% local · 0 data sent to cloud

Real screenshots

See it in action

Scroll to explore →

Chat — main interface

Explain how neural networks learn using backpropagation

✦

Neural networks learn by adjusting weights using the gradient of the loss function. Backprop propagates error backwards through each layer:

loss = cross_entropy(preds, labels)
loss.backward()
optimizer.step() # update weights

Now write me a simple neural net in Python

✦

Sure! Here's a minimal PyTorch network...

Codex — coding agent

Explorer

📄 snake.js

📄 index.html

📁 css/

📄 game.css

📄 README.md

✓ 5 files written

1const canvas = document.getElementById('game');

2const ctx = canvas.getContext('2d');

3// Snake state

4let snake = [{ x: 10, y: 10 }];

5let dir = { x: 1, y: 0 };

6let food = randomFood();

8function tick() {

9 if (!running) return;

10 const head = { x: snake[0].x + dir.x,

11 y: snake[0].y + dir.y };

Aether Media — voice, video, image

On-Demand Media

Aether Media

Upscale images, generate speech, clone voices, create video — all local.

🖼️ Upscale

RealESRGAN 4× boost

Upscale Image

🗣️ TTS

Piper offline TTS

Generate Speech

🎙️ Clone

XTTS voice cloning

Clone Voice

🎬 Video Generation

Describe a clip → ComfyUI renders it locally. Cinematic · Anime · Realistic presets.

Generate Video

Models tab — 108 local models

Model Explorer

Local models, tuned for your PC

Top Picks Coding Vision Reasoning

⚡ Recommended for your system

Based on your GPU. One click to download.

💬 Chat

qwen3:8b

↓ Download

💻 Code

qwen2.5-coder:7b

✓ Active

⚡ Fast

phi4-mini

↓ Download

Generate Image — inline in chat

generate an image of a cyberpunk city at night with neon rain

🖼️ Generate Image FLUX.1-schnell · 1024×1024

🖼️

cyberpunk city at night with neon rain
● Generating... ~10-30s

Describe the image you want to generate...

Cancel

Generate

Complete feature list

All features. All free.

💬

Local AI Chat

Unlimited conversations with 108 local Ollama models. Full history, smart context.

💻

Codex Coding Agent

Opens your folder, reads files, writes code, creates entire projects autonomously.

🎤

Live Voice Assistant

Wake word detection, real-time transcription, spoken replies. Settings → System.

🗣️

Voice Cloning (XTTS)

Clone any voice from a short audio sample. Fully local, multi-language.

🔊

Text to Speech (Piper)

Fast offline TTS using Piper models. No cloud, no latency.

🖼️

Image Generation

FLUX.1 via ComfyUI. Generate images in-chat just by asking. 1024×1024.

🎬

Video Generation

Short local clips via ComfyUI. Cinematic, anime, realistic presets. Seed locking.

🎵

Music & Sound FX

Generate music and sound effects locally using ComfyUI audio nodes.

🔆

Image Upscaling

4× upscale with RealESRGAN via ComfyUI. Fully local.

👁️

Vision & Image Analysis

Drop images into chat. Qwen VL, LLaVA, Moondream can analyse and describe them.

🌐

Web Search

Brave Search integration. Ground answers in real results. 1,000 free/month.

🔍

Deep Check & Compare

Two-model accuracy review. Or compare outputs side by side.

🤖

Custom AI Profiles

Upload docs, set a persona, pick a model. Your private specialist AI.

🎭

Personalities

Persistent AI personas pinned to specific models. Tutor, character, brand voice.

🤝

Autonomous Agents

Multi-step agents with approval gates. Plan, act, iterate across projects.

☁️

Cloud Models

Add OpenAI, Anthropic, or OpenRouter keys. Switch models in one click.

🖥️

Desktop Actions

Let Aether control your PC — launch apps, control media. With approval gates.

📚

Library & Saved Prompts

Save context documents and reusable prompts. Inject into any conversation.

⚡

Auto Model Routing

Automatically routes coding queries to Aether Code, writing to Aether Write.

🗂️

Artifacts & Drafts

AI-generated content saved as editable artifacts. Export to Markdown or JSON.

Prerequisites

Before you download.

A couple of things to have ready. Most people are set up already.

Install Ollama

Aether uses Ollama to run local models. It's a one-click install from ollama.com — takes under a minute. If you already have it, you're good.

✓

ollama.com · free

Check Your RAM

8 GB RAM is the minimum for small models (3–4B). 16 GB gets you the good ones. 32 GB+ and you can run the heavy hitters like DeepSeek R1 70B.

✓

8 GB min · 16 GB rec

GPU = Optional but Fast

An NVIDIA GPU dramatically speeds up inference. AMD and Apple Silicon work too. CPU-only mode works — just slower for large models.

✓

NVIDIA / AMD / Apple Silicon

For Image & Video: Install ComfyUI

Image generation, video, upscaling, and music all require ComfyUI. It's optional — chat and Codex work without it. Setup guide included in the app.

Optional · for media

Free forever · No account required

Download Aether

The full app. Every feature. No subscriptions, no login, no telemetry. Just download and run.

Download for Windows macOS Linux

Windows 10/11 · macOS 12+ · Linux

8 GB RAM minimum

Ollama for local models

NVIDIA GPU recommended