# Langtrace

> Open-source, OpenTelemetry-based observability and evaluations platform for LLM applications, supporting real-time tracing, metrics, and debugging across popular LLMs, frameworks, and vector databases.

Langtrace is an open-source observability and evaluations platform built by Scale3 Labs for teams developing LLM-powered applications and AI agents. It is licensed under AGPL-3.0, with its core application source code fully accessible on GitHub. The platform integrates via a two-line SDK setup in Python or TypeScript and supports both a managed cloud offering and self-hosted deployment.

## What It Is

Langtrace sits in the LLMOps observability category, giving developers end-to-end visibility into their AI stacks. It captures traces, metrics, and logs from LLM API calls, vector database operations, and agent framework interactions using OpenTelemetry (OTEL) standards. The platform surfaces token usage, inference latency, cost, and evaluated accuracy through dashboards, and provides tools for prompt version control, dataset curation, and automated evaluations.

## OpenTelemetry-First Architecture

Langtrace is built on OTEL standards, meaning traces it generates conform to OpenTelemetry semantic conventions. The project maintains a separate repository for semantic span attributes (`langtrace-trace-attributes`) and actively develops conventions for GenAI workloads. The system architecture relies on a Next.js frontend, PostgreSQL as a metadata store, and ClickHouse for storing spans, metrics, logs, and traces.

## Supported Integrations

The platform advertises 30+ integrations across three categories:

- **LLM Providers**: OpenAI, Anthropic, Azure OpenAI, Cohere, DeepSeek, xAI, Groq, Perplexity, Gemini, AWS Bedrock, Mistral
- **LLM Frameworks**: LangChain, LlamaIndex, LangGraph, LiteLLM, DSPy, CrewAI, Ollama, VertexAI, Vercel AI, GuardrailsAI, Agno, Phidata, Cleanlab, and others
- **Vector Databases**: Pinecone, ChromaDB, QDrant, Weaviate, PGVector, MongoDB, Milvus

Both Python and TypeScript SDKs are available, with coverage varying by provider and framework.

## Self-Hosting and Deployment Model

Langtrace explicitly supports self-hosted deployment via Docker Compose, running three services: a Next.js app, a Postgres database, and a ClickHouse database. The project documentation states that self-hosted instances collect no telemetry — no data leaves the user's servers. One-click deployment templates are available for Railway and Northflank. The homepage also notes a new lightweight companion product, Langtrace Lite, described as a fully in-browser OTEL-compatible observability dashboard available at lite.langtrace.ai.

## Update: Version 4.0.11

According to the GitHub repository, the latest release is version 4.0.11, published on April 17, 2025. The repository was last pushed to in November 2025, indicating active ongoing development. The project has accumulated over 1,200 GitHub stars and 125 forks since its creation in March 2024. The repository status badge shows "Active" development.

## Enterprise Security Posture

The platform's homepage states it is SOC 2 Type II certified and uses enterprise-grade encryption standards. Enterprise deployments support custom retention policies, custom SLAs, and on-premises installation options. Security vulnerability reports are directed to security@scale3labs.com.

## Features
- OpenTelemetry-based distributed tracing
- Real-time LLM API call monitoring
- Token usage and cost tracking dashboards
- Inference latency metrics
- Prompt version control and rollback
- Prompt playground for cross-model comparison
- Automated evaluations and dataset curation
- Annotations for human-in-the-loop feedback
- Self-hosting via Docker Compose
- Python and TypeScript SDK support
- SOC 2 Type II compliance
- Enterprise-grade encryption
- 30+ integrations with LLMs, frameworks, and vector DBs
- Langtrace Lite in-browser OTEL dashboard

## Integrations
OpenAI, Anthropic, Azure OpenAI, Cohere, DeepSeek, xAI, Groq, Perplexity, Gemini, AWS Bedrock, Mistral, LangChain, LlamaIndex, LangGraph, LiteLLM, DSPy, CrewAI, Ollama, VertexAI, Vercel AI, GuardrailsAI, Agno, Phidata, Cleanlab, Pinecone, ChromaDB, QDrant, Weaviate, PGVector, MongoDB, Milvus

## Platforms
LINUX, WEB, API, DEVELOPER_SDK, CLI

## Pricing
Open Source, Free tier available

## Version
4.0.11

## Links
- Website: https://www.langtrace.ai
- Documentation: https://docs.langtrace.ai/introduction
- Repository: https://github.com/Scale3-Labs/langtrace
- EveryDev.ai: https://www.everydev.ai/tools/langtrace