Tensormesh Lab is an experimental space for exploring platform capabilities hands-on. Run live demos against serverless models to visualize how KV caching accelerates inference.
Navigate to Tensormesh Lab → Demos from the sidebar to get started.
How It Works
Choose a Demo
Browse the demo catalog and select one to run.
Select a Model
Pick from the available serverless models. No deployment setup required.
Run the Demo
Configure and click Run. The demo streams results in real time.
Available Demos
Ask the Document — Sends 20 questions against a shared long-document prefix to demonstrate KV cache speedup. The first request is cold; follow-ups reuse the cached state and skip the prefill step.
Try the same demo across different models to compare how architecture and model size affect cache speedup.
