Helicone

What it does

Helicone describes itself as an AI Gateway and LLM Observability Platform for AI engineers. The GitHub README lists access to 100+ AI models through an OpenAI-compatible API with intelligent routing and automatic fallbacks, plus quick integration by changing a base URL and adding an API key for providers including OpenAI, Anthropic, Gemini, and frameworks such as LangChain and Vercel AI SDK.

Observability features in the README include inspecting and debugging traces and sessions for agents, chatbots, and document pipelines, analyzing cost, latency, and quality metrics with export to PostHog, a playground for testing prompts and traces, and prompt management with versioning and deployment through the gateway. The platform notes SOC 2 and GDPR compliance for enterprise use and a free tier measured in monthly requests.

The open source stack can be self-hosted with Docker using a docker-compose setup, with a production Helm chart available for enterprise workloads by contact. The architecture comprises a Next.js web frontend, a Cloudflare Workers proxy, a dedicated log collection server, Supabase for application data and auth, ClickHouse for analytics, and Minio for object storage.

Verified capabilities

Quick start

Route a request through the Helicone AI Gateway

Sign up for an API key from the README quick start, point an OpenAI client at the Helicone gateway base URL, and call a model so the request is logged.

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://ai-gateway.helicone.ai",
  apiKey: process.env.HELICONE_API_KEY,
});

const response = await client.chat.completions.create({
  model: "gpt-4o-mini",
  messages: [{ role: "user", content: "Hello!" }]
});

Source: https://github.com/Helicone/helicone. Examples use placeholders only. Never paste a real key into a profile, config file you share, or a ticket.

MCP support: The README notes data can be accessed with an MCP server, and the repository includes a helicone-mcp directory.

Works with

Only sourced support is listed. A missing framework means AgentsUse has not verified it yet, not that it cannot work.

Health and maintenance

Pricing and license

Freemium model with a free Hobby tier including a monthly request allowance and limited storage and retention, paid Pro and Team tiers with usage-based pricing beyond included requests, and custom Enterprise pricing with SSO and on-prem deployment. The pricing page describes simple, predictable pricing that scales with usage.

Official pricing or docs →

Limitations and safety

• Manual self-hosted deployment is explicitly not recommended in the README, which directs users to Docker or Helm instead.
• Production Helm chart for enterprise workloads requires contacting the vendor, rather than a public self-serve download described in the README.
• Free tier is limited by monthly requests and short retention on the pricing page, so high-volume production logging requires a paid usage-based plan.

Alternatives to Helicone

Related tools

Common questions