---
title: "LLM Observability"
description: "Monitor AI applications with Middleware LLM Observability: trace LLM and agent calls, evaluate output quality and monitor NVIDIA GPUs in one place."
section: "LLM Observability"
breadcrumb: "LLM Observability > Overview"
canonical: "https://docs.middleware.io/features/llm"
---
# LLM Observability

LLM Observability makes model calls first-class telemetry. One SDK setup, Middleware's own or OpenLIT and Traceloop over OpenTelemetry, and every call carries its model, prompt, completion, tokens, latency and cost into the same trace as the surrounding request. LLM monitoring charts that per model and per service, LLM evaluation scores the answers with LLM-as-judge and custom evaluators, and GPU monitoring shows the hardware underneath, so a slow response can be traced to a prompt, a provider or a saturated card. The Cookbooks get you from first call to a traced RAG pipeline without reading the spec.

## Features

- [Configure](/features/llm/middleware-sdk) — Add the Middleware, OpenLIT or Traceloop SDK to your app
- [Cookbooks](/features/llm/trace-llm-app) — Worked examples from one model call to a RAG pipeline
- [LLM monitoring](/features/llm/overview) — Tokens, latency, cost and errors per model and per service
- [LLM evaluation](/features/llm/evaluations) — Score responses with LLM-as-judge and your own evaluators
- [GPU monitoring](/features/llm/gpu-monitoring) — Utilization, memory and temperature of your GPUs

## Related Features

- [APM](/features/apm) — Traces, services, profiles and endpoints
- [Infrastructure](/features/infrastructure) — Hosts, Containers, Kubernetes etc.
- [Dashboards](/features/dashboards) — Visualise telemetry data
- [Alerts](/features/alerts) — Create alert rules and service level objectives