helicone  by Helicone

LLM observability platform for monitoring, evaluating, and experimenting

created 2 years ago
4,253 stars

Top 11.7% on sourcepulse

GitHubView on GitHub
Project Summary

Helicone provides an open-source platform for LLM observability, enabling developers to monitor, evaluate, and experiment with their AI applications. It targets developers building with large language models, offering a unified solution to track costs, latency, and quality, manage prompts, and integrate with various LLM providers and frameworks.

How It Works

Helicone acts as a proxy and logging layer, intercepting LLM requests. It supports numerous integrations via a simple code modification or header addition, routing requests through its platform. The core architecture includes a web frontend, a Cloudflare Worker for proxy logging, a dedicated server (Jawn) for log collection, Supabase for application data, ClickHouse for analytics, and Minio for object storage. This distributed setup allows for scalable data ingestion and analysis.

Quick Start & Requirements

  • Install/Run: Self-hosting via docker compose up after cloning the repository and configuring .env. Cloud offering available with a free tier.
  • Prerequisites: Docker for self-hosting. Cloud offering requires API key signup.
  • Links: Docs, Discord, Roadmap, Demo.

Highlighted Details

  • One-line code integration for over 20 LLM providers and frameworks (OpenAI, Anthropic, LangChain, etc.).
  • Features include prompt management, automated evaluations (Ragas, LastMile), fine-tuning partnerships, and a gateway for caching and rate limiting.
  • Cloud offering boasts low latency (~10ms) via Cloudflare Workers.
  • SOC 2 and GDPR compliant for enterprise readiness.

Maintenance & Community

The project is actively maintained, with a public roadmap and a Discord community for support and contributions. Contributions for documentation, integrations, and feature requests are welcomed.

Licensing & Compatibility

Licensed under the Apache v2.0 License, permitting commercial use and integration with closed-source applications.

Limitations & Caveats

Manual self-hosting is explicitly not recommended. The README mentions potential out-of-date integration lists, suggesting direct contact for missing providers.

Health Check
Last commit

12 hours ago

Responsiveness

1 week

Pull Requests (30d)
209
Issues (30d)
50
Star History
567 stars in the last 90 days

Explore Similar Projects

Feedback? Help us improve.