Generative AI Observability, now with GPU monitoring
Documentationbreadcrumb arrow Grafana Cloudbreadcrumb arrow What's new in Grafana Cloudbreadcrumb arrow Generative AI Observability, now with GPU monitoring
Grafana Cloud
← Back to What's new

Generative AI Observability, now with GPU monitoring

Generally AvailableMachine learningIntegrations
Release date: 2025-02-06

We’re excited to unveil our latest update to the Generative AI Observability Cloud integration.

AI Observability is a Grafana Cloud integration we released late last year, designed to provide insights into gen AI application performance.

By leveraging OpenLIT, the OpenTelemetry-native, open-source SDK, it simplifies the monitoring, diagnosis, and optimization of generative AI systems. This integration automatically instruments over 50 gen AI tools, including LLMs, vector databases (vector DBs), and frameworks like LangChain and LlamaIndex, streamlining your setup process.

Our integration not only guides you through the setup but also offers pre-built dashboards that can be customized to fit your needs. These dashboards, focused on LLM & VectorDB Observability, now include an exciting new feature: OpenTelemetry-based GPU monitoring! This new capability enables you to track GPU performance through key metrics such as utilization percentage, temperature, power consumption, and more, allowing you to optimize the efficiency of your AI workloads.

Check out the docs, and let us know what you’d like to see next.