Get Started¶
Choose the shortest path for the result you need. TeleFuser requires a CUDA-capable system for model execution; documentation builds and CPU unit tests do not require a GPU.
Core Path¶
- Install TeleFuser and verify the command-line interface.
- Verify the WebRTC path with the self-contained, model-free preflight.
- Run LingBot-World v2 from the same guide to validate interactive world-model generation in the browser.
For a smaller single-GPU package and batch-API check, use Basic Inference.
Choose a Workflow¶
| Goal | Start here |
|---|---|
| Experience interactive LingBot-World v2 in a browser | Core WebRTC Experience |
| Run a model directly from Python | Basic Inference |
| Expose batch image or video generation over HTTP | Serving and APIs |
| Design or operate a streaming deployment | Stream Server |
| Select a checkpoint and example | Supported Models |
| Diagnose installation or runtime failures | Troubleshooting |
| Integrate a new pipeline | Developer Guide |
Scope¶
TeleFuser covers offline generation, batch HTTP serving, and stateful streaming. The WebRTC preflight separates transport failures from model and GPU failures, while the basic inference path isolates package and batch-service setup. Choose the path that matches the capability you need to validate.