Introduction
Models, pricing, the API, and setup for every supported coding agent.
Zro is a fast, private, multi-region inference service for open models. It exposes the Completions / Responses API and the Messages API, with zero request retention and no training on customer requests.
This is the working reference. Start with the quickstart to make your first request, then dig into the models, the API, or a specific integration.
Start here
Quickstart
Create an API key and send your first request.
Models
Context windows, capabilities, pricing, and upstream open weights.
API reference
Endpoints, streaming, tool calling, reasoning, caching, and regions.
Integrations
Claude Code, Codex, OpenCode, Cursor, Cline, Hermes, and more.
Plans and billing
Spend budgets, usage packs, tool pricing, and per-key controls.
Privacy
Zero request retention, no training on your requests, and regional processing.
Model licenses
Upstream license terms for every model.
At a glance
| Completions / Responses base URL | https://zro.moonmath.ai/v1 |
| Messages base URL | https://zro.moonmath.ai |
| Authentication | Authorization: Bearer sk-... |
| Recommended model | deepseek-v4.1-flash |
| Regions | Europe, United States |
| Tool pricing | See Tool pricing |
Ask the assistant any question using the Ask AI button in the corner. It runs on the same inference endpoint.