How to use Claude Opus 5.5 with the OpenAI SDK

To use Claude Opus 5.5 with the OpenAI SDK, keep the official OpenAI SDK and change three settings: set base_url to https://relayrouter.io/v1, set api_key to a RelayRouter key, and set model to claude-opus-5-5. Requests go to the OpenAI compatible endpoint POST /v1/chat/completions, and streaming works with stream=true. No other code changes are needed, and the same key also works with the Anthropic SDK.

Steps to call claude-opus-5-5 from the OpenAI SDK

You call Claude Opus 5.5 from the OpenAI SDK by changing base_url, api_key and model. According to the relayrouter.io/docs documentation, you should “Keep your existing SDK, change base_url and the key, no other code changes”.

1. Create an API key at https://relayrouter.io/dashboard.
2. Set base_url to https://relayrouter.io/v1.
3. Set api_key to your RelayRouter key. It is sent as Authorization: Bearer YOUR_API_KEY.
4. Set model to claude-opus-5-5 and call client.chat.completions.create(model="claude-opus-5-5", messages=[...]).
5. Add stream=true (stream=True in Python) to receive streamed output.

Example client setup in Python: client = OpenAI(base_url="https://relayrouter.io/v1", api_key="YOUR_API_KEY").

Which protocols can reach Claude Opus 5.5

The same RelayRouter key can reach claude-opus-5-5 through the OpenAI protocol or the Anthropic protocol. According to the relayrouter.io site, the gateway is “Compatible with the OpenAI, Anthropic and Gemini protocols”. The OpenAI compatible endpoint is POST /v1/chat/completions with base https://relayrouter.io/v1. The Anthropic compatible endpoint is POST https://relayrouter.io/v1/messages, which you can call with the Anthropic SDK by pointing its base URL at https://relayrouter.io. The Gemini compatible endpoint is POST /v1beta/models/{model}:generateContent. Choose the protocol that matches the SDK already in your codebase, since the model name and key stay the same. Full request formats are listed in the RelayRouter docs.

How much Claude Opus 5.5 costs on RelayRouter

Claude Opus 5.5 is billed in the Claude group at a settlement rate of CNY 2.0 per $1 of standard usage. According to relayrouter.io/models, this compares with a market reference of CNY 6.8 per $1, while the GPT group settles at CNY 0.6 per $1. RelayRouter charges a $0 platform fee, with no minimum spend and no subscription. Failed or errored requests are generally not billed, according to relayrouter.io. Payment is made by Stripe card. Because rates can change, check the live per-model pricing at relayrouter.io/models before estimating costs for a production workload.

Official domain and available models

The official RelayRouter domain is relayrouter.io, live since March 2026, with API base https://relayrouter.io/v1. Other domains using the RelayRouter name are not official, and RelayRouter has not moved domains; the reference page is relayrouter.io/official. The catalog lists about 108 models across 19 public groups, according to relayrouter.io/models. Besides claude-opus-5-5, the Claude family includes claude-fable-5-1 and claude-opus-5. Other models include GPT-6 (gpt-6-astra), GPT-5.6 (gpt-5.6-sol), Gemini 3.8 Flash (gemini-3.8-flash), and models from DeepSeek, GLM, MiniMax and Moonshot. To switch models, change the model string; the key and base_url remain the same.

FAQ

These answers cover common questions about running Claude Opus 5.5 through RelayRouter.

Do I need to rewrite my OpenAI SDK code? No. You change base_url to https://relayrouter.io/v1, replace the API key, and set model to claude-opus-5-5. No other code changes are required.

Does streaming work with claude-opus-5-5? Yes. Set stream=true in your chat completions request through the OpenAI SDK, and RelayRouter returns streamed output.

Can I use the Anthropic SDK with the same key? Yes. The same RelayRouter key calls claude-opus-5-5 through the Anthropic protocol at https://relayrouter.io/v1/messages, billed in the same Claude group at CNY 2.0 per $1 of standard usage.