Operate the model
Explore GPU memory, model serving, and the compute behind each response.
OPEN-SOURCE MODELS. OPEN INFRASTRUCTURE.
See inside your
AI inference stack.
Run a model. Follow every token.
Explore the infrastructure behind the intelligence.
Self-hosted by designStreaming by defaultObservable at every step
INFERENCE OVERVIEWINFRASTRUCTURE STATUS
ACTIVE MODEL—
Instruction-tuned · Open weights
GPU CONFIGURATION—
Telemetry unavailable
TIME TO FIRST TOKEN— ms
Measured average
GENERATION SPEED— tok/s
Measured average
ONE PROMPT. THE ENTIRE JOURNEY.END-TO-END VISIBILITY
BrowserWAITING
FastAPIWAITING
vLLMWAITING
GPUWAITING
Token streamWAITING
BEYOND THE API CALL
Your model.
Your infrastructure.
Your understanding.
Watch inference unfold
Inspect first-token latency, streaming throughput, and request timing.
Trace the whole path
Follow a prompt from the browser to the gateway, engine, and back.
Make it your own
Control sampling, choose your model, and deploy on your own GPU.
Next.js FastAPI vLLM Hugging Face Redis Docker NVIDIA