Current release: v0.0.6 — new decide + embed services, Gemini 3.8 Flash + TTS, Claude Haiku 5.5, Nano Banana 2.1, FLUX 3 Image, Qwen-Image-2.1, FASHN VTON 1.5 (local), Veo 3.1 / Omni 1.1 / Seedance references, Cosmos 3 local, ElevenLabs TTS, DeepSeek (132 models, 9 services).

📚 Full documentation: https://tryonlabs.github.io/opentryon/
API tutorials, configuration, examples, and agent guides live there — not in this README.

What you get

Four ways to use it

1. CLI — opentryon <service> --model <model> [params...]
2. MCP server — expose every registry model as tools for Claude, Cursor, or TryOn Studio. Guide: MCP Server · mcp-server/README.md
3. TryOn Studio — Next.js UI (Agent, Connect, Image, VTON, Understand, Video, BG Remove) over MCP HTTP. Setup: TryOn Studio
4. Python — from tryon.api import ... (and tryon.cli.runner.invoke_model)

Install

git clone https://github.com/tryonlabs/opentryon.git
cd opentryon
conda env create -f environment.yml
conda activate opentryon
pip install -e .
# Optional local/GPU models: pip install -e ".[local]"

Or with pip: pip install -r requirements.txt && pip install -e .

cp env.template .env   # add the API keys you need

Details: Installation · Configuration

Quick start

# Dry-run (no API call) — verifies CLI + registry wiring
opentryon vton --model flux-vto \
  --person-image data/model-1.jpg --garment-image data/garment.png --dry-run

# Real call (needs BFL_API_KEY in .env)
opentryon vton --model flux-vto \
  --person-image data/model-1.jpg --garment-image data/garment.png \
  --garment-description "olive green bomber jacket"

# Multimodal understanding (needs MOONSHOT_API_KEY)
opentryon understand --model kimi-k3 --image data/model-1.jpg \
  --prompt "Describe this outfit for a product listing." --reasoning-effort high

from tryon.api import KimiUnderstandAdapter

adapter = KimiUnderstandAdapter(model="kimi-k3")
result = adapter.understand_image(
    "data/model-1.jpg",
    prompt="Describe this outfit.",
    reasoning_effort="high",
)
print(result["text"])

More examples: Quickstart · CLI · API Reference

CLI services

opentryon <service> --model <model> [params...]
opentryon understand --help                    # list models
opentryon understand --model kimi-k3 --help    # list that model's flags

Models marked local need pip install opentryon[local]. Full table and flags: Unified CLI.

MCP server

cd mcp-server
pip install -r requirements.txt
python server.py                                    # stdio (Claude Desktop / Cursor)
python server.py --transport http --host 127.0.0.1 --port 8000   # TryOn Studio

Tools are generated from tryon/cli/registry.py — the same registry as the CLI. Guide: MCP Server · full tool table: mcp-server/README.md. Web UI: TryOn Studio.

Demos & notebooks

This package ships Gradio demos and Jupyter notebooks only:

python run_demo.py --name extract_garment   # also: model_swap, outfit_generator

Notebooks: notebooks/. Web UI: TryOn Studio.

Layout

opentryon/
├── tryon/           # Package: api/, cli/, models/, agents/, datasets/, preprocessing/
├── tryondiffusion/  # Research diffusion training / inference
├── mcp-server/      # FastMCP server (registry → tools)
├── openapi/         # OpenAPI / Swagger snapshot (upstream media APIs)
├── postman/         # Postman collection for media providers
├── demo/            # Gradio demos
├── notebooks/       # Jupyter examples
├── docs/            # Docusaurus documentation site
├── tests/           # CLI / adapter smoke checks
└── env.template     # API key template

Documentation map

When contributing docs or APIs: put long tutorials in docs/, not this README. Keep README as the project front door only.

Contributing

See CONTRIBUTING.md. Open an issue before large changes; prefer PRs that update the registry, tests, and docs together.

License

Creative Commons BY-NC 4.0. Non-commercial use with attribution to this repository; indicate any changes you make.

Star History

Made with ❤️ by TryOn Labs · Discord