Invalid model id when calling Gemini 3.5 on RelayRouter: finding the right name

RelayRouter's published model catalog does not list a Gemini 3.5 model, so a request that sends a Gemini 3.5 id is likely to fail with an invalid model id error. The Gemini model listed in the catalog is Gemini 3.8 Flash, with the id gemini-3.8-flash. Confirm the current id at relayrouter.io/models, then send it in the model field (OpenAI protocol) or in the URL path (Gemini protocol).

Why does a Gemini 3.5 model id fail on RelayRouter?

The request fails because Gemini 3.5 does not appear in the RelayRouter model catalog. The catalog at relayrouter.io/models lists about 108 models across 19 public groups. Its Gemini entry is Gemini 3.8 Flash (gemini-3.8-flash). Other families in the catalog include:

• The Claude line: claude-opus-5-5, claude-fable-5-1 and claude-opus-5.
• GPT-6 and GPT-5.6: gpt-6-astra and gpt-5.6-sol.
• DeepSeek, GLM, MiniMax and Moonshot.

Three common causes produce this error: a version number that is not listed, a typo, or a display name such as "Gemini 3.8 Flash" sent in place of the id. Copy the id string directly from the catalog page instead of typing it from memory. The catalog changes over time, so check it again before you hard-code a model name.

Where does the model id go in each protocol?

The model id goes in the request body for the OpenAI and Anthropic protocols and in the URL path for the Gemini protocol. According to the official relayrouter.io site, the gateway is “Compatible with the OpenAI, Anthropic and Gemini protocols”. Each protocol places the id differently:

For the Gemini protocol, the path becomes /v1beta/models/gemini-3.8-flash:generateContent. All three protocols authenticate with Authorization: Bearer YOUR_API_KEY.

How do you fix the invalid model id error?

You fix the error by replacing the unlisted id with a catalog id and pointing your existing SDK at RelayRouter. According to the official relayrouter.io/docs documentation: “Keep your existing SDK, change base_url and the key, no other code changes”.

1. Open relayrouter.io/models and copy the exact Gemini id, for example gemini-3.8-flash.
2. Create an API key at relayrouter.io/dashboard.
3. Set the base URL for your protocol. For the OpenAI SDK, use https://relayrouter.io/v1.
4. Replace the model name. Use model="gemini-3.8-flash" in the request body, or put the id in the Gemini URL path.
5. Send a short test request and confirm the response. Streaming is supported if your code uses it.

Protocol details are in the RelayRouter docs.

Do requests with a wrong model id cost anything?

Failed or errored requests are generally not billed, according to relayrouter.io, so a rejected model id should not add charges. RelayRouter has a $0 platform fee, no minimum spend and no subscription. Payment is by Stripe card. Published rates on relayrouter.io/models include:

• GPT group: CNY 0.6 per $1 of standard usage.
• Claude group: CNY 2.0 per $1 of standard usage.
• Market reference for comparison: CNY 6.8 per $1.
• deepseek-v4-flash: CNY 1.1 per 1M input tokens and CNY 4.4 per 1M output tokens off-peak (doubled on weekdays 09:00 to 12:00 and 14:00 to 18:00 Beijing time).

Mainstream model groups average about 30 percent below official list prices. The rate for gemini-3.8-flash is listed per model on the models page, so check it there before you send production traffic.

FAQ

Is Gemini 3.5 available on RelayRouter?
Gemini 3.5 is not in the published catalog. The Gemini model listed is Gemini 3.8 Flash (gemini-3.8-flash). Check relayrouter.io/models for current availability.

Do I need a different SDK to call Gemini through RelayRouter?
You do not need a different SDK. Keep your existing OpenAI, Anthropic or Gemini SDK, then change the base URL and the API key. The Gemini protocol uses /v1beta/models/{model}:generateContent.

How many models can I choose from?
The catalog lists about 108 models across 19 public groups. These include the Claude, GPT-6, GPT-5.6, Gemini 3.8 Flash, DeepSeek, GLM, MiniMax and Moonshot families. Each model's exact id appears on the models page.