Skip to content

Cascade v0.1.0

Pre-release
Pre-release

Choose a tag to compare

@abhee235 abhee235 released this 07 Oct 17:03
· 12 commits to main since this release

The first public release of Cascade — an open-source, local-first AI app builder. Describe the app you
want; an AI coding agent writes the code, runs it and shows you a live preview. It works with your own
Ollama models: no cloud, no account.

This is an early pre-release. Please open an issue
if anything goes wrong.

Download

System File How to install
Windows 10/11 (64-bit) CascadeSetup.exe Run it. Cascade installs and opens.
macOS (Apple Silicon) Cascade-darwin-arm64-0.1.0.zip Unzip it and drag Cascade.app to Applications.
Linux (x64) Cascade-linux-x64-0.1.0.zip Unzip it and run ./cascade.

The first launch shows a warning, because this build is not code-signed yet:

  • Windows: "Windows protected your PC" → click More info → Run anyway.
  • macOS: right-click Cascade.app → Open → Open.

What you also need

  • A model. Either install Ollama and pull a tool-capable model
    (ollama pull qwen3:8b), or add your own API key for OpenAI, Groq, OpenRouter, NVIDIA or any
    OpenAI-compatible server.
  • Memory. 8 GB of RAM works with a hosted model. 16 GB or more is recommended for local models, ideally
    with a GPU.
  • Docker is optional. By default, projects run directly on your computer. For full isolation per
    project, install Docker Desktop and switch the
    runtime to Docker in Settings.

What's in it

  • Build from a prompt: chat, a live preview at desktop, tablet or phone size, a code editor, a terminal,
    and a saved version for every change that you can restore.
  • Any model: local Ollama models or your own API keys. Set the context window, output length and
    sampling for each model.
  • A real agent: tools, permissions, sub-agents, memory and context compaction, tuned to get whole apps out
    of local models.
  • Connectors (MCP): web search, live library docs and more, or add any MCP server.
  • Observatory: every turn is traced locally: each model call, tool call and token count.

Updates

On Windows, Cascade checks this Releases page and installs new versions automatically. On macOS and Linux,
download new versions from here until builds are signed.

Build it yourself

Prefer running from source? The README has the
commands.