Glossary term
Glossary term
Infrastructure and Serving
LLMOps refers to the practices and tools used to deploy, manage, and monitor large language models in production. It covers the full lifecycle, including data preparation, model tuning, deployment, evaluation, and ongoing optimization. LLMOps ensures that AI systems remain reliable, scalable, and compliant while maintaining performance over time.
LangSmith by LangChain, Helicone, and Langfuse are LLMOps platforms for tracing and evaluation.
Arize Phoenix, Datadog LLM Observability, and Weights & Biases Weave offer LLMOps capabilities.
TruEra and Fiddler AI provide LLMOps for model evaluation, drift, and governance.