Lunary

Lunary is an observability and evaluation platform for applications built on large language models (LLMs). It helps engineering and product teams see how AI features behave in production and quickly pinpoint issues.

Observability and analytics for LLMs

Lunary collects request logs, model outputs, and key quality metrics so you can understand what’s happening across your AI workflows.

• Track performance, reliability, and cost over time
• Analyze user conversations with chatbots
• Identify where responses fail, drift, or miss expectations

Prompt management and experimentation

Lunary includes tools to manage prompts as a workflow, making it easier to improve quality without ad-hoc code changes.

• Prompt versioning and comparisons
• A/B testing and evaluations
• Structured iteration on prompt wording and behavior

Built for startups and enterprise

Lunary can be used for internal AI tools as well as customer-facing products. Teams get a clearer view of LLM behavior and data to support product decisions instead of treating the model like a black box.