emu — AI For The Rest Of Us

emu is a local-first, multimodal AI desktop application that runs state-of-the-art AI models entirely on your own computer. Chat with text, images and audio; dictate by voice; have replies read aloud; and extract text from documents — all with a clean, minimal interface, and with no cloud account, no telemetry, and no network calls for inference.

A product of CloudKites AI Lab (CloudKites Pty Ltd), Sydney, Australia.

💚 Free to download, use, and share. emu is distributed free of charge. You may use it and redistribute the complete, unmodified installer for free (no selling, no paid bundling). See License below.

🎉 Latest release: version 2.0.0 — a ground-up "zero-config" redesign. See What's new in 2.0.0 below.

⬇️ Download — which installer do I need?

emu 2.0.0 ships for Windows 10/11 (64-bit) and Linux (x86-64), each in a CPU and a GPU (CUDA) edition — identical in features, differing only in the hardware they use. Pick one from the table (or the Files tab).

Your computer	Edition	Installer	Size
Windows + NVIDIA GPU (GTX 16-series / RTX 20-series or newer)	GPU (CUDA)	emu-2.0.0-windows-cuda-setup.exe	~758 MB
Windows, no NVIDIA GPU (integrated / AMD / Intel graphics)	CPU	emu-2.0.0-windows-cpu-setup.exe	~62 MB
Linux + NVIDIA GPU (GTX 16-series / RTX 20-series or newer)	GPU (CUDA)	emu-2.0.0-linux-cuda.run	~866 MB
Linux, no NVIDIA GPU (integrated / AMD / Intel graphics)	CPU	emu-2.0.0-linux-cpu.run	~119 MB

One quick question decides it: do you have an NVIDIA GPU with Compute Capability 7.5+ (GTX 16-series / RTX 20-series or newer) and a current NVIDIA driver (see requirements below)? Yes → the CUDA edition (fastest; the GPU runtime is bundled, nothing extra to install). No → the CPU edition (runs on any modern PC; smallest download). AMD and Intel graphics are not CUDA GPUs — choose CPU.

Not sure? The CPU edition runs everywhere — including on machines with a GPU — so it's always a safe choice; you can install the CUDA edition later.

The Windows installers are code-signed by CloudKites Pty Ltd (Microsoft Azure Trusted Signing); the Linux .run installers are served over HTTPS (Linux has no OS-level code-signing equivalent). All editions are self-contained — the GPU editions bundle the NVIDIA CUDA/cuDNN runtime, so there is nothing else to install. AI models are not in the installer; on first run emu offers to download the model bundle(s) you choose into your local models folder, after which everything runs offline.

📄 Need help choosing or installing? See Download & Installation Guide (PDF).

🎉 What's new in 2.0.0

Version 2.0.0 is a ground-up "zero-config" redesign — emu now sets itself up for your machine and puts every capability one click away.

Zero-config setup — emu detects your CPU, memory and GPU and automatically picks and downloads a chat model that runs well on your hardware. No manual model hunting; sensible defaults out of the box.
One-tap, unified workspace — a cleaner single-window layout with one button per task: attach an Image, run OCR, dictate (🎤), and speak replies (🔊). Settings are organized and simplified.
Dedicated image understanding — a specialized vision model reads and answers questions about your images (identify objects, read signs and labels, describe scenes), in addition to the general multimodal chat.
Larger, smarter chat — support for a new flagship chat model on capable GPUs, plus faster on-GPU generation.
Better documents — OCR now shows live progress, handles multi-page PDFs, and every extraction is saved to a document manager you can reopen and export.
Voice you can keep — spoken (text-to-speech) clips are saved for replay and export; right-click any text to Speak it; language-aware voice selection.
Smoother & more robust — many stability, performance, and quality fixes across chat, speech, and OCR.

Everything still runs 100% on your device — no account, no telemetry, no cloud inference.

🖥️ Minimum requirements

Component	Minimum	Recommended
Operating system	Windows 10/11 (64-bit) or Linux (x86-64, modern distro)	Windows 11 / a current Linux distro
Processor	64-bit CPU with AVX2 (Intel 2013+ / AMD 2015+ — virtually any modern CPU)	Recent multi-core CPU
Memory	16 GB RAM	32 GB RAM
Free disk	8 GB (app + first model)	20 GB+
GPU (CUDA editions only)	NVIDIA, Compute Capability 7.5+ (Turing / GTX 16-series, RTX 20-series or newer), 6–8 GB VRAM	Compute Capability 8.6+ (RTX 30-series or newer), 12 GB+ VRAM
NVIDIA driver (CUDA editions only)	Driver supporting CUDA 13 — Windows release 595.xx or newer (Linux R595+); simplest is the latest Game Ready / Studio driver	Latest available
Internet	Once, to download model bundles on first use	—

CUDA edition — what you need (and don't). You only need an NVIDIA driver; no CUDA Toolkit, cuDNN, or developer tools are required — the GPU runtime is bundled and emu ships prebuilt native GPU code for a wide range of NVIDIA GPUs (Turing → Blackwell), so nothing is compiled on your machine. Run nvidia-smi and check that the CUDA Version shown is 13.x; if it is lower, update your driver. If the GPU or driver is unsupported, emu falls back to the CPU automatically.

🛠️ Install

Windows: double-click the .exe. If SmartScreen appears, confirm the publisher reads CLOUDKITES PTY LTD, choose Run, accept the licence, and finish. Shortcuts are created in the Start menu and on the Desktop.

Linux: make the download executable and run it — chmod +x emu-2.0.0-linux-*.run && ./emu-2.0.0-linux-*.run — then follow the installer wizard. It adds an application-menu entry; emu is self-contained (Qt and, for the CUDA edition, the NVIDIA runtime are bundled — you only need an NVIDIA driver). Microphone dictation uses your system recorder (PipeWire/PulseAudio/ALSA), present on any desktop Linux.

On first run, open the model picker and let emu download a chat model suited to your hardware.

✨ Using emu

Multimodal chat — type in your language; attach an image (with a thumbnail preview — OCR it or chat about it) or an audio clip.
🎤 Speech-to-text — dictate prompts on-device (auto-detect or pick a language).
🔊 Text-to-speech — read replies aloud; multiple engines, including on-device Vietnamese voice cloning. Right-click any text to Speak it, and spoken clips are saved for replay/export.
🔊 Replay/save spoken audio — replay the last spoken (text-to-speech) clip, or export it to a WAV.
📄 Document OCR — /ocr (or the image preview's OCR button) an image or PDF; extracted text and tables drop into the chat.
Hardware-aware model picker — emu detects your CPU/RAM/GPU and offers the models that run well on your machine, defaulting to the best fit.
Persistent chats, Markdown rendering, and resource-aware loading. Settings cover device (CPU/CUDA/Auto), context length, and more.
100% private & offline — every model runs on your CPU or NVIDIA GPU; nothing leaves your device.

English and Vietnamese are first-class across chat, speech, and voice; the chat models natively handle dozens more languages.

📚 Documentation

Read these for full details (also on the Files tab):

Download & Installation Guide (PDF) — choosing an edition, downloading, installing, and getting started.
User Guide (PDF) — complete, step-by-step guide to every feature.
Product Information (PDF) — full product overview: capabilities, bundled AI models and their licences, system requirements, privacy, and limitations.

📜 License

✅ Free to use — personal or commercial, on any number of devices.
✅ Free to redistribute — share the complete, unmodified official installer, free of charge, with all notices intact.
❌ No selling / no paid bundling, no modification, no reverse-engineering.
The source code of emu and its Numbat ML toolkit is confidential and is not distributed.

"emu", "Numbat" and "CloudKites" are trademarks of CloudKites Pty Ltd.

License documents (all on the Files tab):

End-User License Agreement — LICENSE (plain text) — the full terms, presented during installation.
End-User License Agreement — LICENSE.pdf (formatted PDF) — the same terms.
Third-party attributions — THIRD_PARTY_LICENSES.md — licences for the bundled AI models and libraries (Gemma, Qwen, gwen-tts, Pocket TTS, VieNeu-TTS, PP-OCRv6, Qt, NVIDIA CUDA/cuDNN), each used under its own terms.

⚠️ AI output notice

emu uses generative AI models. Output may be inaccurate, incomplete, or fabricated even when it appears confident, and is not a substitute for professional advice — verify important information against trusted sources.

💬 Support & contact

Questions or issues? We're happy to help.

Email: contact@cloudkites.com
Website: https://cloudkites.com/contact

Downloads last month: -; Downloads are not tracked for this model. How to track