mozilla-ai/otari

What it solves

Otari solves the complexity of managing multiple LLM providers and controlling costs. Instead of integrating various APIs directly into your application, you use a single endpoint that handles authentication, routes requests to different providers, and enforces budgets before any spend occurs.

How it works

Otari acts as a proxy server between your applications and LLM providers (like OpenAI or Anthropic). It provides an OpenAI-compatible interface, meaning your existing client code can point to Otari instead of a provider. It uses a routing engine to connect to over 40 providers, manages virtual API keys for your users, and tracks usage and spending in real-time.

Who it’s for

It is designed for developers and organizations that need to centralize their AI stack, manage API credentials securely without exposing them to client applications, and implement strict cost controls and usage monitoring across multiple LLM models.

Highlights

• Unified Interface: One endpoint that supports OpenAI and Anthropic API formats across 40+ providers.
• Cost Control: Enforces per-user and per-key budgets before requests are processed.
• Secure Credential Management: Keeps provider keys in a controlled environment, issuing scoped virtual keys to clients.
• Observability: Provides built-in usage tracking and an admin dashboard for monitoring spend and models.
• Extensible Features: Supports built-in tools like sandboxed Python code execution and web search, as well as request-level guardrails.