SegFault42/HeliosGen

🏆 Self Hosted Gen AI desktop app. Free alternative to Higgsfield, OpenArt, Freepik....

HeliosGen – Visual AI Image & Video Workflow Builder

What it is – HeliosGen is an open‑source desktop application that lets you design, run, and reuse AI‑powered image‑ and video‑generation pipelines on an infinite node‑based canvas. All processing happens locally; the app simply forwards requests to the kie.ai API (or optionally to a locally‑installed Codex CLI) and stores results on your machine.

Why it matters – Instead of typing a single prompt at a time, you can chain together multiple prompts, models, reference images, and automation steps. This makes it easy to experiment with multi‑model workflows, create reusable assets, and keep full control over your data and credits.

Quick start

Download the pre‑built macOS package from the Releases page (Windows/Linux builds are awaiting contributors).
Open the app, go to Settings → API Keys, and paste a kie.ai API key (you obtain one at https://kie.ai).
Begin building a workflow: drag nodes onto the canvas, connect them, select the desired model, and hit Run.
All generated media, workflow definitions, and app settings are saved in a local folder (e.g., ~/Library/Application Support/cash.sdd.helios.desktop/ on macOS). Deleting that folder resets the app.

Core features

Infinite node‑based canvas – visually arrange prompts, models, reference images, and automation steps.
Multi‑model pipelines – mix and match any of the supported image or video models within a single workflow.
Reference image support – feed existing images into generation steps.
Parallel & sequential execution – run branches concurrently or in order.
Real‑time history – see each generation as it completes.
Local‑only operation – no cloud service, no accounts; only the external AI API is contacted.
Optional Codex CLI backend – use your own ChatGPT Codex subscription for GPT‑Image‑2 generation.
Cross‑platform desktop shell – built with Tauri 2, so the UI runs on macOS, Windows and Linux (builds for the latter two are community‑driven).

Supported models (as of the README)

Images – GPT Image 2 (OpenAI), Nano Banana family (Google), Seedream 5.0, Z‑Image, Grok Imagine, etc.

Videos – Veo 3.1 series (Google), Kling series, Seedance, Grok Imagine video, HappyHorse (Alibaba), and more to come.

Tech stack

Layer	Technology
Desktop shell	Tauri 2 (Rust)
UI & logic	Next.js + React + TypeScript (bundled Node sidecar)
Persistence	SQLite (local file)
Media storage	Local disk
AI backend	kie.ai (REST API)

Building yourself

If you prefer to compile the app or need a Windows/Linux build:

git clone https://github.com/SegFault42/HeliosGen
cd HeliosGen
npm install          # installs Node deps
npm run desktop:build # produces OS‑specific bundles

You need Node 22+, Rust, and the platform‑specific Tauri prerequisites (Xcode tools on macOS, C++ Build Tools + WebView2 on Windows, webkit2gtk‑4.1 etc. on Linux).

License

HeliosGen is released under the MIT License, allowing free use, modification, and distribution.

Who might benefit?

Creators who want a visual, reusable way to experiment with AI image/video generation.
Developers building custom AI pipelines who prefer a local, no‑cloud UI.
Anyone looking to keep full ownership of their generation credits and data.

Built for creators building the future of AI workflows.