Backed by Combinator
Fully open-source

Observability Platform for LLM-developers

Get a demoStart Building
1from openai import OpenAI
2

3client = OpenAI(
4  api_key=OPENAI_API_KEY,
5  base_url="http://oai.hconeai.com/v1", 
6  default_headers= { 
7    "Helicone-Auth": f"Bearer {HELICONE_API_KEY}",
8  }
9)

Trusted by thousands of companies

Built to scale with your business

  • Any Model

    We support models from OpenAI, Claude, Gemini, and more.

  • Any Scale

    Log millions of requests per second with no latency impact.

  • Async Packages

    We offer async packages for all major languages and frameworks.

  • Fully Open-Source

    Deploy Helicone on-prem for maximum security.

One observability platform, everything you need

Collect data, monitor performance, and improve your LLM-powered application over time

App screenshot
Labels and Feedback.
Easily segment requests, environments, and more with custom properties.
Caching.
Lower costs and improve performance as well as configure your cache responses.
User Rate Limiting.
Rate limit power users by requests, costs, and more to prevent abuse.
Alerts.
Get notified when your application is down, slow, or experiencing issues.
Key Vault
Securely map and manage your API keys, tokens, and other secrets.
Exporting.
Extract, transform, and load your data using our GraphQL API, webhooks, and more.

Enterprise

Get to production-quality faster

Helicone makes it easy for companies to innovate faster and smarter, ensuring your team can stay ahead of the competition.

    Experiments

    Experiments

    Test prompt changes and analyze results with ease. Easily version prompts and compare results across datasets

    • Prompt Regression Testing
    • Version and Test
    • Dataset Comparison and Analysis
    Schedule a Demo
    Customer Portal

    Customer Portal

    Share Helicone dashboards with your team and clients. Easily manage permissions and access.

    • Share Dashboards
    • Manage Rate-Limits and Permissions
    • Customize Branding and Look and Feel
    Schedule a Demo
    Fine-Tuning

    Fine-Tuning

    Collect feedback and improve model performance over time. We make it easy to fine-tune and export data.

    • Fine-tune OpenAI Models
    • Collect and label data for fine-tuning
    • Export Data as CSV or JSONL
    Get Started

Developer

Fully Open-Source

We believe in the power of community and the importance of transparency. Helicone is fully open-source and available for anyone to use.

Our Tech Stack

  • Frontend: React, Next.js, TailwindCSS
  • Backend: Supabase, Clickhouse, Postgres, Node, Express
  • Infrastructure: Cloudflare, AWS, Vercel

Frequently asked questions