Nebius Inference Observability

Access real-time and historical metrics (latency, throughput, errors, scaling) for your Nebius Token Factory inference workloads via Prometheus-compatible API.

Prerequisites

pip install requests pyyaml
export NEBIUS_API_KEY="your-static-api-key"    # create in UI: Project settings → API keys
export NEBIUS_PROJECT_ID="your-project-id"     # Project settings → Copy project ID

Metrics endpoint

https://monitoring.tokenfactory.nebius.com/v1/projects/{PROJECT_ID}/prometheus/metrics

All requests require Authorization: Bearer {API_KEY}.