GPT text model pricing has been updated. All GPT text models are now billed at 10% of OpenAI's official API prices, including the latest official price reductions for GPT-5.6 Terra and Luna.View latest pricing

AI Gateway · Reliability first

One API. Every top model, only success is billed.

Route text, image and video workloads through a single OpenAI‑compatible endpoint. Multi-layer redundancy keeps production up, and any unused balance can be withdrawn at any time — no subscription, no seats.

One API for leading AI models and providers

OpenAIOpenAIClaudeAnthropicGoogleGoogleSoonMetaMetaSoonDeepSeekDeepSeekSoonMistralMistralSoonGrokxAISoonQwenQwenSoonMinimaxMiniMaxSoonByteDanceByteDanceSoonKimiMoonshotSoonZhipuZ.aiSoonGroqGroqSoonNvidiaNVIDIASoonPerplexityPerplexitySoon

Callable today: OpenAI and Anthropic. The rest are rolling out — the live list is whatever GET /v1/models returns for your key.

01Every Model

One API for Unified Access to Leading AI Models

RouteMux provides enterprises with a unified gateway to AI model access. With a single integration, teams can flexibly connect to leading models across providers, reducing duplicated development and maintenance effort while making model adoption, switching, and scaling significantly more efficient.

  • OpenAI
  • Anthropic
  • Google
  • ROUTEMUX BEST ROUTE
02Every Route

Multi-Layer Redundancy Architecture for High-Availability AI Services

Stability is not an optional feature. It is a core principle embedded in RouteMux's architecture from the ground up. By building a comprehensive, fully redundant system across the network, gateway, and model service layers, RouteMux delivers greater resilience, failover capability, and recovery performance under abnormal conditions, minimizing the impact of interruptions on production workloads.

  • Unified billingCredits shared by every model family
  • Redundant routingHealth-aware routing across vendors
  • Refundable balanceWithdraw any unused balance, anytime

One account, one invoice, one place to watch it all.

03Every Dollar

Worry-Free Refunds, Greater Confidence in Use

RouteMux offers a worry-free refund policy for greater peace of mind. Users may request a refund at any time for any unused account balance, reducing the burden of locked-in funds while providing greater flexibility and certainty throughout the service experience. With transparent policies and a more user-friendly approach, RouteMux helps businesses adopt, test, and scale AI with greater confidence.

Not just access. Every request tightens cost, routing and visibility.

LLMS.TXT

One prompt — use RouteMux from any Agent

Paste this into Claude Code, Codex, Cursor or any agent and it learns RouteMux's base URL, model names and billing rules on its own — no docs reading required first.

routemux — llms.txt prompt

Please read RouteMux's documentation index at https://routemux.com/llms.txt, then answer my question. About RouteMux: an AI gateway exposing hundreds of models through OpenAI / Anthropic / Vertex compatible protocols, billed from a prepaid wallet, charged only for successful requests, with a refundable balance. My question: how do I call its OpenAI-compatible endpoint from Python?

MODELS

Curated top models across categories — start building in minutes.

View all models

Connect in three steps

Within minutes you can start calling hundreds of AI models.

  1. Create an API key

    Sign up, top up your wallet, and mint a key in the console.

  2. Point base_url at RouteMux

    Keep your OpenAI or Anthropic SDK — only the base_url changes.

  3. Call any model

    One key and one balance for hundreds of models across providers.

routemux — openai/gpt-5.6-sol
curl https://api.routemux.com/v1/chat/completions \
  -H "Authorization: Bearer $ROUTEMUX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-5.6-sol",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

WHY ROUTEMUX

Why teams choose RouteMux

One platform, clear pricing, real support — so you can focus on shipping.

01

Official channels

Named vendor routes with stated SLAs instead of opaque resale.

02

Transparent pricing

Per-token and per-call numbers you can check line by line and put in a budget.

03

One console

Keys, quotas, usage and logs in one place — no tab-hopping between vendor dashboards.

04

Fast integration

OpenAI / Anthropic / Vertex compatible. Change one base_url and ship the same day.

05

Refundable balance

Request a refund for any unused balance at any time. Your money isn't locked in.

06

Humans on call

When an integration or invoice needs attention, an engineer answers — not a ticket bot.

FAQ

Frequently asked questions

What is an AI gateway (AI router)?

An AI gateway is a unified access layer between your application and model providers. RouteMux puts hundreds of AI models behind one API and one key, handling routing, billing, and failover on the backend so you don't integrate each provider separately.

How is RouteMux different from OpenRouter, and is it an alternative?

RouteMux is an alternative to aggregator gateways like OpenRouter, built reliability-first: multi-layer redundancy for high availability, transparent usage-based billing, and a refundable balance.

How do I use RouteMux with cc switch, Claude Code, or Cursor?

Point the tool's base_url at RouteMux's gateway and use your RouteMux key. OpenAI, Anthropic, and Vertex protocols are supported, so existing setups switch over by changing one line.

Which models and protocols are supported?

Hundreds of models across OpenAI, Anthropic, Google, DeepSeek, MiniMax, ByteDance and more, compatible with the OpenAI, Anthropic, and Vertex protocols. New models are added continuously.

How does billing work, and can I get a refund?

Usage-based, prepaid credits with one unified bill and transparent markup, no monthly minimum. Any unused account balance can be refunded at any time.

Am I charged for failed calls, and what are the rate limits?

Only successful requests are billed — failures and timeouts cost nothing. Rate limits are configured per account and the current RPM/TPM and concurrency ceilings are visible in the console, not hidden.

GET STARTED

One key to connect. Start calling.

Sign up for $0.5 in trial credit — about 25M input tokens.

Trusted by developers · Powered by leading model providers