How much does Agent Aleph cost?

Agent Aleph is completely free and open source (MIT license). There are no subscriptions, API keys or per-token costs.

v0.2.0-beta · Linux x64 · Free

Your code.
Your models.
Your machine.

AI coding agent and model manager, 100% local.
No subscriptions. No cloud. Your data never leaves your computer.

Download .AppImage Download .deb View on GitHub

100% private Works offline CPU & GPU (CUDA / Vulkan) Free and open source

qwen3.6-35b · active

qwen3.6-35b-a3b-instruct-q4_k_m

Refactor the authentication module to use native JWT

reasoning

I need to review the module structure, identify its external dependencies and design a native HMAC-SHA256 implementation…

response

I'll implement JWT with native crypto.subtle.

Type your message…

Tools · 3/7

read_file

src/auth/jwt.ts

write_file

src/auth/jwt.ts

run_command

npm test auth

llama-server · port 8080

4.8 tok/s · ctx 2048

Local chat Coding agent Model manager 100% private

Download

Built on open-source technology

llama.cpp Tauri 2 Rust Svelte 5 Hugging Face GGUF OpenAI API compat. CUDA Vulkan llama.cpp Tauri 2 Rust Svelte 5 Hugging Face GGUF OpenAI API compat. CUDA Vulkan

Everything you need,
without depending on anyone

Model manager + autonomous coding agent, running on your hardware.

100% private

Your code, prompts and conversations never leave your machine. No telemetry, no remote logs, no API keys.

Works offline

Install once, download the model and you're set. No external servers, no outages, no network latency.

Hugging Face Hub

Explore thousands of GGUF models by topic: code, reasoning, agents and more.

Hardware detection

Per-model badge: whether it fits in VRAM, spills to RAM, or won't run on your machine.

CPU and GPU

Runs on any x64 CPU. With a CUDA or Vulkan GPU it accelerates automatically.

Total privacy

The AI that
won't spy on you

Every model runs on your machine. Every token is generated locally. None of your data — code, prompts or conversations — ever leaves your computer.

No telemetry or remote logging

No API key, no account, no subscription

Works without an internet connection

data in the cloud

∞

tokens, no limit

cost per query

v0.2.0-beta · Available now

Get started in minutes

Download, install and have your first model running in under 5 minutes.

.AppImage — portable .deb — Debian/Ubuntu

🐧 Linux x64 · 💾 8 GB RAM minimum · 🖥 GPU optional · 📦 ~500 MB install

FAQ

Frequently asked questions

Questions? If you can't find what you're looking for, use the contact form.

Write to us

Does Agent Aleph send my data to the cloud?

No. Agent Aleph is 100% local. Every model runs on your machine and no data — code, prompts or conversations — leaves your computer. No telemetry, no remote logs.

Do I need a GPU to use it?

No. It runs on any x64 CPU with 8 GB of RAM. If you have an NVIDIA (CUDA) or Vulkan-compatible GPU, performance improves automatically. Agent Aleph shows you which models fit your hardware before downloading them.

Which models can I use?

Any model in GGUF format available on the Hugging Face Hub: Qwen, Mistral, DeepSeek, Llama, Phi, Gemma and many more. There's a curated catalog by topic (code, reasoning, agents) with a hardware-compatibility badge.

How much does it cost?

Completely free. Open source under the MIT license. No subscriptions, no API keys, no per-token cost. Download once and it's yours forever.

Which operating systems does it run on?

Currently only Linux x64 (.AppImage and .deb). Windows and macOS support is planned. If you'd like to help with the porting, the repository is open.

Contact

Let's talk

Found a bug? Want to contribute? Have a proposal? Write to us directly.

Bug report

Something isn't working as expected

Contribution

You want to add a feature or do porting

General inquiry

Questions, feedback or ideas

You can also open an issue on GitHub

Your code. Your models. Your machine.

Everything you need,without depending on anyone