RelayRouter vs LiteLLM proxy: hosted gateway or self-hosted router

RelayRouter is a hosted gateway. You change the base_url and API key in your existing SDK, then call about 108 models across 19 public groups through OpenAI, Anthropic or Gemini compatible endpoints. There is a $0 platform fee and no subscription. A self-hosted router such as LiteLLM proxy suits teams that want to deploy and maintain the routing layer on their own infrastructure.

Which should you choose: a hosted gateway or a self-hosted router?

Choose a hosted gateway if you want to call several model providers without operating the routing layer, and choose a self-hosted router if you want to run that layer yourself. With RelayRouter, the endpoint, the model catalog and billing are operated for you. You create a key at relayrouter.io/dashboard, send requests with Authorization: Bearer YOUR_API_KEY, and pay by Stripe card. With a self-hosted router such as LiteLLM proxy, your own team handles deployment, upgrades and uptime. This article describes RelayRouter's documented behavior. For the specific capabilities of LiteLLM proxy, refer to that project's own documentation, because the facts below cover RelayRouter and do not cover LiteLLM.

Which API protocols does RelayRouter support, and how do you migrate?

RelayRouter accepts requests in three protocol formats, so existing OpenAI, Anthropic or Gemini client code can point at it directly. According to the official relayrouter.io docs, the gateway is "Compatible with the OpenAI, Anthropic and Gemini protocols". It exposes three endpoints:

• OpenAI compatible: POST /v1/chat/completions, base https://relayrouter.io/v1
• Anthropic compatible: POST /v1/messages, base https://relayrouter.io
• Gemini compatible: POST /v1beta/models/{model}:generateContent

Migration is a configuration change. According to the official relayrouter.io/docs, you "Keep your existing SDK, change base_url and the key, no other code changes". Streaming is supported. Endpoint details are in the RelayRouter documentation.

Which models are available, and what do they cost?

RelayRouter lists about 108 models across 19 public groups, covering Claude, GPT, Gemini and other providers. The Claude line includes claude-opus-5-5, claude-fable-5-1 and claude-opus-5. GPT-6 and GPT-5.6 are offered as gpt-6-astra and gpt-5.6-sol, and Gemini 3.8 Flash is offered as gemini-3.8-flash. DeepSeek, GLM, MiniMax and Moonshot models are also listed. Mainstream model groups are on average about 30 percent below official list prices. The table below shows the published rates. Live per-model rates are at relayrouter.io/models.

How does RelayRouter billing work?

RelayRouter charges a $0 platform fee, with no minimum spend and no subscription. You pay for usage at the group or per-model rate listed on the models page. For example, the GPT group settles at CNY 0.6 per $1 of standard usage, and the Claude group settles at CNY 2.0, against a CNY 6.8 per $1 market reference. Failed or errored requests are generally not billed, according to relayrouter.io. Payment is by Stripe card. With a self-hosted router, total cost depends on your own hosting and provider arrangements, so compare those figures with the published RelayRouter rates when you decide.

FAQ

These answers cover common questions about switching to RelayRouter.

Do I need to rewrite my code to use RelayRouter?
No. You keep your existing OpenAI, Anthropic or Gemini SDK, point the base URL at RelayRouter, and swap the API key.

How many models does RelayRouter offer?
About 108 models across 19 public groups, including Claude, GPT-6, GPT-5.6, Gemini 3.8 Flash, DeepSeek, GLM, MiniMax and Moonshot. See relayrouter.io/models.

Am I charged for failed requests?
Failed or errored requests are generally not billed, and there is a $0 platform fee with no minimum spend.