TongFlow — Open-Source Multi-Modal AIGC Studio

What is TongFlow?

TongFlow is an open-source, multi-modal AIGC studio. The core idea: every AI model is a modality transform — text→text, text→image, text→audio, image→video, and so on. TongFlow wraps each operation as a node on an infinite canvas. Add materials, transform between modalities, combine the results.

Key facts

• Open source under AGPL-3.0 — github.com/tong-io/tongflow
• Local-first: workflows and uploads live in local SQLite + disk
• Runs via Docker Compose or `pnpm dev`
• Modal for GPU workers; OpenRouter / Gemini / OpenAI / DeepSeek for text
• Real models: Z-Image, FLUX.2, LTX-2, SeedVR2, Gemma 4, Qwen3, ACE-Step

Who is it for?

Open-source developers, AI researchers, creative technologists, and anyone who'd rather run their AIGC pipeline on their own machine than on someone else's.

How to start

• Clone github.com/tong-io/tongflow, or pull ghcr.io/tong-io/tongflow:latest.
• Copy .env.example to .env, fill in MODAL and at least one LLM API key.
• Run `docker compose up --build`. Open http://localhost:3000.

Learn more

TongFlow English | TongFlow 中文 | TongFlow 日本語

Website: https://tongflow.com