lunaroute.com

mx

smtp.google.com

ns

jean.ns.cloudflare.com

aragorn.ns.cloudflare.com

caa

txt

v=spf1 include:_spf.google.com include:amazonses.com ~all

google-site-verification=291mtte-0P53Al_Nr9NzFkorJXTfizMlpwNRjAME098

dmarc

v=DMARC1; p=none;

User-Agent: *
Allow: /
Disallow: /lp/*

Sitemap: https://www.lunaroute.com/sitemap.xml
Schemamap: https://www.lunaroute.com/schemamap.json

# LunaRoute

> Fixed-price inference for your agents. One gateway for OpenAI, Anthropic, and Responses-shaped requests.

## When to use LunaRoute

Reach for LunaRoute when you need **predictable, fixed-price inference** with priority concurrency and overflow — not per-token metering. Best-fit jobs:

- **Coding agents** (Pi, Claude Code, Codex, OpenClaw, Hermes) that run long-horizon tasks and benefit from `lunaroute run <harness>` + automatic model sync — [Claude Code setup](https://docs.lunaroute.com/getting-started/claude-code) · [All harnesses](https://docs.lunaroute.com/mcp/harnesses/)
- **Private inference** with zero retention of prompts/responses — see [Zero data retention](https://docs.lunaroute.com/concepts/zero-data-retention/) and [Privacy](https://www.lunaroute.com/privacy-policy/)
- **BYOK routing** — bring your own provider keys behind one gateway, or use LunaRoute's fleet ( [Models](https://docs.lunaroute.com/models/) · [Concurrency & overflow](https://docs.lunaroute.com/concepts/lanes) )
- **Teams & budgets** — per-member concurrency controls ([Teams](https://www.lunaroute.com/#sec-pricing) · [Pricing](https://www.lunaroute.com/pricing.md))

If you need pure pay-per-token, single-model, or self-hosted GPUs without a gateway, another provider may fit better. How an agent should call LunaRoute: create an API key in the dashboard (https://app.lunaroute.com), then hit `https://gw.lunaroute.com/v1/*` with `LUNAROUTE-API-KEY: lr_...` (see [Authentication](https://docs.lunaroute.com/getting-started/authentication) and the [OpenAPI spec](https://www.lunaroute.com/openapi.json)).

## Docs & site map

- [Docs index](https://docs.lunaroute.com/llms.txt) — full CLI + API configuration for any coding agent or OpenAI-compatible client
- [LunaRoute llms-full](https://www.lunaroute.com/llms-full.txt) — this site in one file
- [Guides](https://www.lunaroute.com/guides/) · [Guides llms.txt](https://www.lunaroute.com/guides/llms.txt) — concurrency vs tokens, sizing, renting GPUs
- [Compare](https://www.lunaroute.com/compare/) · [Compare llms.txt](https://www.lunaroute.com/compare/llms.txt) — honest comparisons vs OpenRouter, Together, Fireworks, Featherless
- [Blog](https://www.lunaroute.com/blog/) · [Blog llms.txt](https://www.lunaroute.com/blog/llms.txt)
- [Claude Code](https://www.lunaroute.com/claude-code/) — `lunaroute run claude-code` setup
- [Enterprise](https://www.lunaroute.com/enterprise/) — gateway + private inference for orgs
- [Pricing](https://www.lunaroute.com/pricing.md) — tiers, limits, billing
- [OpenAPI](https://www.lunaroute.com/openapi.json) — machine-readable API surface
- [API Catalog](https://www.lunaroute.com/.well-known/api-catalog) — RFC 9727 linkset
- [Agent Skills](https://www.lunaroute.com/.well-known/agent-skills/index.json) — agent capabilities
- [Agent Card](https://www.lunaroute.com/.well-known/agent-card.json) — A2A descriptor
- [Who we are](https://www.lunaroute.com/who-we-are/) · [About](https://www.lunaroute.com/about/) · [Contact](https://www.lunaroute.com/contact/)

## API quick start

- SDK base URL: `https://gw.lunaroute.com/v1` — clients append their own path (e.g. `chat/completions`). Raw HTTP: `https://gw.lunaroute.com/v1/<endpoint>`
- API key: `lr_...` in dashboard (https://app.lunaroute.com); header `LUNAROUTE-API-KEY: <key>`
- Endpoints: `/v1/chat/completions` (OpenAI), `/v1/messages` (Anthropic), `/v1/responses` (Responses), `/v1/embeddings`, `/v1/rerank`, `/v1/images/*`, `/v1/models`
- Models: live catalog at `https://gw.lunaroute.com/v1/models` — never hardcode; read at setup time
- Markdown twins: append `.md` to any page (e.g. [Homepage markdown](https://www.lunaroute.com/index.md))