OpenLIT is an open-source AI engineering platform focused on native OpenTelemetry observability. With a single line of code, it instruments your LLM applications and provides full-stack monitoring covering language models, vector databases, and GPUs.

OpenLIT auto-instruments more than 50 providers of LLMs, AI frameworks, and vector databases, with each integration producing OpenTelemetry traces and metrics. An analytical dashboard tracks the health and performance of your applications, costs, and user interactions. The platform also provides guardrails, evaluations, prompt management, a Vault for your secrets, and a Playground.

Designed to be vendor-neutral, OpenLIT exports its data to any OTLP-compatible backend (Grafana Tempo, Datadog, Honeycomb, raw OpenTelemetry collector…). It is the ideal tool to measure, understand, and control your costs and the quality of your AI applications, on infrastructure managed by DINAO.