Helicone

Helicone is an LLM observability and AI gateway platform with 5.8k+ GitHub stars — a LangSmith alternative acquired by Mintlify in March 2026. One-line integration for logging, cost tracking, and prompt management.

Key features

Observability & logging

• One-line integration with OpenAI, Anthropic, Cohere, and more
• Request/response logging with full prompt capture
• Cost tracking per request, model, and user
• Latency monitoring and performance analytics
• Custom properties and user tracking

AI gateway

• Proxy-based architecture — zero code changes required
• Rate limiting and usage quotas
• Load balancing across API keys and providers
• Caching for repeated prompts
• Fallback chains for reliability

Prompt management

• Version control for prompts
• A/B testing and experiments
• Prompt playground for debugging
• Template management and sharing

Analytics & insights

• Real-time dashboards and alerts
• User-level usage analytics
• Model comparison and benchmarking
• Error tracking and debugging
• Export to Datadog, BigQuery, and S3

At a glance

License	Apache-2.0
Stack	TypeScript, Python, PostgreSQL
Self-hosted	Yes — Helicone OSS
Cloud	helicone.ai (managed)
Integration	Proxy-based, one-line

Self-hosting

git clone https://github.com/Helicone/helicone.git
cd helicone && docker compose up

Helicone can be self-hosted with Docker. The managed cloud provides automatic updates and scaling.