# RouterPlus

> One API key for many LLM providers. OpenAI-compatible /v1/chat/completions and
> Anthropic-compatible /v1/messages on one gateway. Pass-through pricing,
> per-request cost on the wire, honest failover that never switches mid-answer.

Gateway base URL: https://api.routerplus.com
Create an account in your browser: https://app.routerplus.com/signup. Complete secure verification, then create an API key at https://app.routerplus.com/console/keys.

## Start here
- [Agent integration guide](https://app.routerplus.com/docs/agent-integration.md): read this first to build the gateway into a product — keys, model calls, provider pinning, sessions and caching, errors, limits and cost
- [All docs in one file](https://app.routerplus.com/llms-full.txt): every page below as raw markdown, in this order

## Get started
- [Overview](https://app.routerplus.com/docs/overview.md): what this is, both wire surfaces, and your first call
- [Quickstart](https://app.routerplus.com/docs/quickstart.md): signup to streamed call in five steps
- [Agent integration guide](https://app.routerplus.com/docs/agent-integration.md): one page for an AI agent that builds the gateway into a product: keys, model calls, provider pinning, sessions and caching, errors, limits and cost
- [Playground](https://app.routerplus.com/docs/playground.md): chat, decision, images and video with any catalog model; what it costs and what is stored
- [Optimize](https://app.routerplus.com/docs/model-search.md): find a cheaper model or mix that scores as well as yours on your Braintrust evals, and deploy it
- [Authentication](https://app.routerplus.com/docs/authentication.md): keys, both header styles, and secure browser signup
- [Bring your own key](https://app.routerplus.com/docs/byok.md): provider connections, validation, rotation, and free metered inference
- [Routing policies](https://app.routerplus.com/docs/routing-policies.md): ordered connections, explicit funding fallback, and Azure
- [Workspaces and principals](https://app.routerplus.com/docs/identity.md): membership, scoped keys and trusted end-user quotas
- [Bedrock identity](https://app.routerplus.com/docs/bedrock.md): AWS workload roles, regional Claude models and signed streaming
- [Limits and capacity](https://app.routerplus.com/docs/admission.md): shared quotas, provider pools, fairness, and outage behavior
- [Dedicated endpoints](https://app.routerplus.com/docs/dedicated-endpoints.md): capacity reserved for your organization on one model: its own limits, contract prices and same-model failover

## Core concepts
- [Models & catalog](https://app.routerplus.com/docs/models.md): the catalog, model ids, uptime honesty, machine-readable feeds
- [Pricing & billing](https://app.routerplus.com/docs/pricing.md): pass-through prices, usage.cost on the wire, the billing math
- [Streaming](https://app.routerplus.com/docs/streaming.md): SSE on both surfaces, keep-alives, cancellation semantics
- [Errors](https://app.routerplus.com/docs/errors.md): every error class, per-surface shapes, exact remediation
- [Routing & failover](https://app.routerplus.com/docs/routing.md): the commit boundary, no silent substitution, health circuits
- [Rate limits & spend caps](https://app.routerplus.com/docs/limits.md): per-key RPM, organization token and concurrency limits, monthly caps, the 429 with a reset time
- [Data policy](https://app.routerplus.com/docs/data-policy.md): content-free by design; what is and is not stored

## API reference
- [POST /v1/chat/completions](https://app.routerplus.com/docs/api-chat-completions.md): the OpenAI-compatible surface
- [POST /v1/messages](https://app.routerplus.com/docs/api-messages.md): the Anthropic-compatible surface
- [POST /v1/images/generations](https://app.routerplus.com/docs/api-images.md): OpenAI Images API: base64 in the body, all-or-nothing
- [POST /v1/videos](https://app.routerplus.com/docs/api-videos.md): OpenAI Videos API: a job, then the file; billed at the provider's cost
- [POST /v1/decisions](https://app.routerplus.com/docs/api-decisions.md): typed answers from a decision model: Jev from TypeSafe, Mercury Decide from Inception, Bespoke Nimble v3 from Bespoke Labs, Clef and Clef-flash from Cloudflare, Decider 2B or Kev 4B from RouterPlus (our own), Perplexity Decider v1 27B from Perplexity, Sage from Levanto or GLiNER-2.5-Decide from Fastino, each in its own question format (eight of them share System One's)
- [GET /v1/models](https://app.routerplus.com/docs/api-models.md): the catalog, in whichever shape your SDK expects
- [GET /v1/usage & /v1/generation](https://app.routerplus.com/docs/api-usage.md): spend, balances, and the per-attempt audit trail
- [Retired signup endpoint](https://app.routerplus.com/docs/api-signup.md): create accounts through secure browser signup

## Guides
- [Migrate](https://app.routerplus.com/docs/migration.md): from OpenAI, Anthropic, or OpenRouter — copy-paste diffs
- [Coding agents](https://app.routerplus.com/docs/install.md): Claude Code, Codex, Cursor, aider — wire any harness to the gateway
- [Wire compatibility](https://app.routerplus.com/docs/compat.md): the D8 conventions your bill is computed from

## For providers
- [List your models](https://app.routerplus.com/docs/for-providers.md): one manifest, a published conformance bar for text, image and video, paid prepaid or in arrears

## Machine-readable
- [Model catalog JSON](https://app.routerplus.com/api/models.json): ids, prices, context windows
- [Provider manifest schema](https://app.routerplus.com/docs/provider-manifest.schema.json): list your models (supply side)