Aqueduct¶
Aqueduct is TU Wien's API gateway for AI services. It provides large language models, text-to-speech, speech-to-text, image generation and MCP servers through OpenAI-compatible endpoints. You can use it from your own code, coding agents and editors, or any other tool that speaks the OpenAI API.
All models run on TU Wien hardware. Aqueduct is available to researchers, lecturers and university staff.
-
Get an API key and send your first request.
-
Language, speech and image models with their capabilities.
-
Set up Pi, OpenCode, Crush or Zed with Aqueduct.
-
Give your AI tools access to TU Wien resources.
Links¶
| Resource | Link |
|---|---|
| Aqueduct UI (API keys, usage, teams) | aqueduct.ai.datalab.tuwien.ac.at |
| API base URL | https://aqueduct.ai.datalab.tuwien.ac.at/v1 |
| Support | #llm-service:tuwien.ac.at on Matrix |
| Status | Monitoring dashboard |
| Usage examples | LLM-as-a-Service cookbook |
| API reference | Public Aqueduct documentation |
| Source code | Aqueduct on GitHub, contributions welcome |