TouchDesigner component

TD OpenRouter

300+ AI models in one TouchDesigner COMP.

A single drop-in component that connects your project to models from OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, Black Forest Labs and more: text, image, video, audio, embeddings and transcription, through one API key.

Get it on Gumroad Features
td-openrouter.toe — /project1
The TD_OpenRouter component in TouchDesigner, wired to status, text, audio, image, video, transcription and embedding outputs
300+
AI models
7
modalities, one COMP
1
API key
0
Python packages to install

Features

  • 01

    Every modality

    Text with streaming and multi-turn history, image generation, async video generation, text-to-speech and audio, embeddings, and speech-to-text. Each on its own parameter page.

  • 02

    Discover, then pick

    One pulse fetches the live model list and fills every dropdown, sorted free-first and tagged for vision and video input, with context length.

  • 03

    Vision and video in

    Send a TOP, a local file or a URL alongside your prompt. Gemini even takes YouTube links directly.

  • 04

    Results as operators

    Text lands in DATs, images and videos load into TOPs, audio plays through a CHOP, and embeddings arrive as data you can wire anywhere.

  • 05

    Never blocks the timeline

    Requests run on background threads with an activity log that timestamps every event. Abort a generation mid-stream and keep what you have.

  • 06

    Stdlib only

    No external dependencies. It uses the Python that ships with TouchDesigner, so it works on any machine you drop it on.

Outputs

Drop it in. Wire it up.

Every output is a regular TouchDesigner operator, ready to wire into the rest of your network.

BaseTD_OpenRouterstatustext_responseimage_resultvideo_resultaudio_pathtranscription_resultembeddings_result

Built for live work

Whether you're building a live AV performance, an interactive installation, a generative art piece or a real-time prototype, TD OpenRouter lets you tap into the latest frontier models without writing a line of HTTP code or installing a single Python package.

Quick start

  1. Drag TD_OpenRouter.tox into any network.
  2. Paste your OpenRouter API key on the Connection page.
  3. Pulse Discover Models to fill every model dropdown.
  4. Switch to Text Generation, type a prompt, pulse Generate.

Parameter pages

PageWhat it does
ConnectionAPI key, sampling (temperature, top P, top K, max tokens), Discover Models
Text GenerationChat with optional image and video references, JSON mode, streaming
Image GenerationText-to-image with size and output folder
Video GenerationText-to-video, submitted and polled in the background
EmbeddingsText to vector
Audio GenerationText-to-speech or audio generation, routed automatically
TranscriptionSpeech-to-text