# RouterBase

> A unified OpenAI-compatible API that routes requests to 200+ AI models from OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, xAI, and more through a single key.

RouterBase is a unified LLM routing platform operated by DeepOpen, LLC that lets developers access 200+ frontier AI models through a single OpenAI-compatible API endpoint. It covers chat, image, video, and audio modalities from providers including OpenAI, Anthropic, Google DeepMind, Meta, Mistral, DeepSeek, xAI, Cohere, Stability AI, Runway, ElevenLabs, and Black Forest Labs. The homepage states the platform is trusted by 10,000+ developers and has served 2B+ tokens.

## What It Is

RouterBase sits between your application and the upstream AI providers, acting as a routing and reliability layer. Instead of maintaining separate SDKs, API keys, billing accounts, and fallback logic for each provider, developers point the OpenAI SDK at a single RouterBase base URL and gain access to the full model catalog. The core value proposition is eliminating integration overhead: one key, one invoice, one observability dashboard.

## How the Routing Layer Works

RouterBase exposes three routing behaviors that operate per request:

- **Smart routing** — automatically selects the best provider for a given request based on price, latency, or availability signals.
- **Automatic fallback** — switches to an alternate provider in milliseconds when the primary returns an error or times out, targeting zero-downtime behavior.
- **BYOK (Bring Your Own Key)** — supported for OpenAI, Anthropic, and Google; token costs pass through to the user's own provider account while the platform fee remains flat.

## OpenAI-Compatible Drop-In

The API follows the OpenAI wire format for chat completions, embeddings, and image endpoints. Developers swap one base URL (`routerbase.com/v1`) and keep existing code unchanged. According to the documentation, tools like LangChain, LlamaIndex, the Vercel AI SDK, Cursor, and Continue work through their existing OpenAI provider configuration without modification. Official SDKs are available for Python, Node.js, Go, and Ruby.

## Model Catalog and Modalities

The pricing page lists 84 total models at time of crawl, spanning 58 chat, 8 image, and 17 video models. Providers represented include OpenAI (GPT-4o, GPT-4.1, GPT-5 series), Anthropic (Claude Haiku, Sonnet, Opus lines), Google (Gemini 2.5 and 3.x series), DeepSeek (R1, V3, V4 lines), ZAI (GLM-4.x and GLM-5.x), MoonshotAI (Kimi K2), and others. Image and video providers include Flux, Ideogram, Veo, Kling, and Runway. Audio is covered by ElevenLabs and Suno.

## Observability and Compliance Features

Every API call is logged with tokens, latency, cost, and full payload. Logging is opt-in (off by default) for privacy. Upstream providers receive requests with zero-retention flags set where supported. RouterBase does not use prompts to train any models. EU, US, and APAC regions are available; Enterprise plans support per-key region locks and signed DPAs. Per-key rate limits are configurable in the dashboard, with near-cap webhooks that fire before users hit a 429.

## Pricing Model and Volume Tiers

RouterBase uses a pay-as-you-go credit model with pass-through token pricing plus a flat platform fee. The pricing page shows per-model discounts ranging from 5% to 15% off provider list prices. New accounts receive $5 in free credits without requiring a credit card. The platform states that accounts spending above $1k/month automatically qualify for Silver, Gold, and Ultra volume tiers with progressively deeper discounts. Enterprise contracts are available for teams needing bespoke SLAs, reserved capacity, private deployments, and committed-use discounts.

## Features
- OpenAI-compatible API (drop-in replacement)
- 200+ AI models via single API key
- Smart routing by price, latency, or availability
- Automatic provider fallback in milliseconds
- Unified billing across all providers
- Real-time usage analytics and spend tracking
- Streaming SSE support
- Tool calling and function calling
- Vision inputs
- JSON mode
- Prompt caching
- BYOK (Bring Your Own Key) for OpenAI, Anthropic, Google
- Per-key rate limit configuration
- Near-cap webhooks
- EU/US/APAC region selection
- Privacy-first logging (opt-in)
- Zero-retention flags on upstream requests
- CSV export and API query for usage data
- Playground for testing models before coding
- Auto top-up and low-balance alerts

## Integrations
OpenAI SDK, LangChain, LlamaIndex, Vercel AI SDK, Cursor, Continue, Python SDK, Node.js SDK, Go SDK, Ruby SDK, OpenAI, Anthropic, Google DeepMind, Meta, Mistral, DeepSeek, xAI, Cohere, Stability AI, Runway, ElevenLabs, Suno, Black Forest Labs, Flux, Ideogram, Veo, Kling

## Platforms
WEB, API

## Pricing
Freemium — Free tier available with paid upgrades

## Links
- Website: https://routerbase.com
- Documentation: https://docs.routerbase.com/
- EveryDev.ai: https://www.everydev.ai/tools/routerbase