# Helicone

> Helicone provides observability and analytics for large language model usage via a web dashboard and API to capture telemetry, metrics, and logs from LLM calls.

Helicone provides observability and analytics for large language model (LLM) usage, capturing telemetry, token usage, latency, and request metadata through a proxy or SDK and exposing results in a web dashboard and API. It helps teams monitor LLM performance, troubleshoot failures, and understand cost-driving usage patterns. Helicone is designed to be deployed self-hosted or used as a hosted service to collect LLM metrics and logs.

- **Telemetry capture** — Use Helicone as a proxy or instrument the SDK to capture requests, responses, tokens, and latency for LLM calls.
- **Dashboard and API analytics** — View aggregated metrics, token consumption, and latency trends in the web dashboard or query metrics programmatically via the API.
- **Request logging and replay** — Store request/response pairs and metadata for debugging and post-hoc analysis.
- **Integrations and SDKs** — Install language- or framework-specific SDKs or route traffic through the Helicone proxy to start collecting data.
- **Self-host or hosted** — Deploy the open-source components to your infrastructure for local data control or use the hosted offering for managed telemetry.

To get started, create an account or deploy the open-source components, configure your LLM client to use Helicone's proxy or SDK, and open the dashboard to inspect metrics and logs. Use the API for automated metric queries and export.

## Features
- LLM observability and analytics
- API proxy and SDK instrumentation
- Token usage and cost analytics
- Latency and performance metrics
- Request/response logging for debugging
- Web dashboard and metrics API

## Integrations
OpenAI, Anthropic, Hugging Face, Custom HTTP-based LLM endpoints

## Platforms
WEB, API, DEVELOPER_SDK

## Pricing
Open Source, Free tier available

## Links
- Website: https://www.helicone.ai
- Documentation: https://docs.helicone.ai/
- Repository: https://github.com/helicone/helicone
- EveryDev.ai: https://www.everydev.ai/tools/helicone