Laminar is an AI-powered platform for engineering LLM products with 3k+ GitHub stars — a LangSmith alternative for agent observability, evaluation, and debugging.
Key features
Observability
- Distributed tracing for LLM chains and agents
- Signal alerts for anomalies and performance issues
- Real-time monitoring of model quality and latency
- Custom metrics and dashboards
- Integration with OpenTelemetry
Evaluation & testing
- Automated evals with custom scorers
- Regression testing for prompt changes
- Side-by-side comparison of model versions
- Human-in-the-loop annotation
- Dataset management for evals
Debugging
- Trace visualization with input/output inspection
- Error tracking and root cause analysis
- Prompt version diffing
- Latency breakdown by component
- Cost attribution per trace
Developer experience
- Python and TypeScript SDKs
- REST API for custom integrations
- Webhook alerts to Slack, Discord, email
- Team collaboration and role-based access
- Self-hosted and cloud options
At a glance
| |
|---|
| License | Apache-2.0 |
| Stack | Python, TypeScript, PostgreSQL |
| Self-hosted | Yes — Laminar OSS |
| Cloud | lmnr.ai (managed) |
| Tracing | OpenTelemetry-compatible |
Self-hosting
git clone https://github.com/lmnr-ai/lmnr.git
Laminar can be self-hosted for full control over your data. The cloud version provides managed infrastructure and automatic updates.