Extensions/Tanvo
ComfyUI Extension

Tanvo

Tanvo nodes for ComfyUI: images on Nano Banana, Seedream, GPT Image and Qwen; video on Kling, Veo, Seedance and Wan; songs on Suno; and over 100 ready-made photo, video and song apps. The free model works without an API key.

By tanvoai·Created about a month ago·Updated a day ago· 0
tanvoai/ComfyUI-Tanvo
Nodes—
On cloudLocal install
Stars0
Updateda day ago
Readme

Tanvo for ComfyUI

license

ComfyUI nodes for Tanvo, an AI image, video and music studio. Generate images on Nano Banana, Seedream, GPT Image, Qwen and Grok Imagine, video on Kling, Veo, Seedance, Wan, LTX and MiniMax (most with sound), and songs on Suno V6. Or run one of Tanvo's 100+ ready-made apps (pet portraits, figurines, old photo restoration, tattoo designs, holiday videos, birthday songs) on your own photos, with the same tuned prompts and settings as the website.

Everything runs on Tanvo's servers. No GPU, no model downloads. The free model works without an account.

Install · API key · Nodes · Example workflows · Pricing · Troubleshooting

Install

ComfyUI Manager: search for Tanvo and install, then restart ComfyUI.

Comfy CLI:

comfy node install tanvo

By hand:

cd ComfyUI/custom_nodes
git clone https://github.com/tanvoai/ComfyUI-Tanvo
pip install -r ComfyUI-Tanvo/requirements.txt

Needs ComfyUI 0.8 or newer. The only dependency is the tanvo Python client (no other packages).

API key

Without a key the nodes use Tanvo's free tier: the free image model (studio-image-v1), watermarked, with a small daily allowance. For every other model:

  1. Sign in at tanvo.ai (new accounts start with free credits).
  2. Create a key under Settings → API keys. It starts with sk_.
  3. In ComfyUI open Settings → Tanvo and paste it. The next run uses it; no restart needed.

The key is kept in ComfyUI's own settings file, never in your workflows, so you can share workflow files and images safely. You can set TANVO_API_KEY in the environment instead; it takes precedence.

Nodes

All nodes are under Tanvo in the node library (double-click the canvas and type "tanvo").

| Node | What it does | Outputs | |---|---|---| | Tanvo Image | Text to image, or edit and combine images. Connect up to 8 image inputs (every image of a batch counts). | images, url | | Tanvo Video | Text to video, or animate a start_image, optionally ending on an end_image. Most models add sound. | video, url | | Tanvo Song | A song in two takes: describe it, sing your own lyrics, or go instrumental. | take_1, take_2, lyrics | | Tanvo Photo App | One of Tanvo's photo apps with its preset looks. Pick the app and the photo inputs it needs appear. | images, url | | Tanvo Video App | Trend video apps (duets, hugs, dances, product spins, video to cartoon). Some take a video clip. | video, url | | Tanvo Song App | Birthday songs, love songs. Put the name and a few details in details. | take_1, take_2, lyrics |

Picking a model shows only that model's settings (aspect, resolution, duration, sound, Veo's quality tier), so you cannot pick a combination it does not support. The url output is the hosted file, handy for passing on to other services.

Seed. Tanvo does not use it. It is there so you can re-run a node on purpose: ComfyUI skips nodes whose inputs have not changed, so queueing the same workflow twice does not pay twice. Change the seed (or set it to increment) to get a new result.

Cancelling stops ComfyUI from waiting. The run itself carries on at Tanvo and lands in your history; failed runs are refunded automatically.

New models and apps appear after restarting ComfyUI: the nodes read Tanvo's catalogue at startup and fall back to the copy they ship with when offline.

Example workflows

In the templates browser they are listed under the folder Tanvo is installed in (tanvo or ComfyUI-Tanvo). The JSON files are also in example_workflows.

  • Text to image (free): Tanvo Image → Save Image. Works without a key.
  • Pet portrait app: Load Image → Tanvo Photo App (Pet Portrait Generator, Royal look) → Save Image.

Pricing

Runs cost credits at the same prices as the website; the pricing page lists every model. Failed runs are refunded. The free model costs nothing.

Troubleshooting

| Message | Fix | |---|---| | needs_api_key, allowance or anon_ip_daily | Add a key under Settings → Tanvo. | | credits | Top up at tanvo.ai/pricing. | | invalid: … options.resolution | That model does not take the setting; pick another value. | | busy | Your plan's parallel runs are in use; wait for one to finish. | | Nodes missing after install | Check the ComfyUI console for errors mentioning Tanvo, and that ComfyUI is 0.8 or newer. |

Development

python -m unittest tests/test_core.py                     # no ComfyUI needed
COMFYUI_PATH=~/ComfyUI ~/ComfyUI/.venv/bin/python -m unittest tests/test_e2e.py

The end-to-end test runs every node in a real ComfyUI against a local stand-in for the Tanvo API (link this folder into custom_nodes first). python scripts/update_catalog.py refreshes the bundled catalogue.

Also from Tanvo

License

MIT