How to generate images with gpt-image and DALL-E 3 on RelayRouter

The provided facts do not document image generation, gpt-image or DALL-E 3 support on RelayRouter, so this article cannot confirm those capabilities. What the facts do confirm: RelayRouter is a unified AI API gateway that is 「Compatible with both the OpenAI and Anthropic protocols」 (据 relayrouter.io 官方文档), exposing /v1/chat/completions and /v1/messages. For current model availability, always check the live list at relayrouter.io/models.

What RelayRouter documents today

RelayRouter documents two API protocols and a defined set of text model families. It speaks both OpenAI (/v1/chat/completions) and Anthropic (/v1/messages) protocols, per relayrouter.io/models. Model coverage includes the Claude family, GPT-5.5, Gemini 3.5, plus DeepSeek, GLM, MiniMax and Moonshot. None of these 4 protocol and coverage facts reference image generation endpoints, gpt-image or DALL-E 3. If image support exists, it would appear on the live models page, which lists per-model rates. Treat that page as the source of truth rather than this article.

How migration works

Migration to RelayRouter requires only configuration changes, not code rewrites. 「Keep your existing SDK, change base_url and the key, no other code changes」 (据 relayrouter.io/docs 官方文档). This means you point your current OpenAI or Anthropic SDK at RelayRouter and swap the API key, then continue calling the same 2 protocols you already use. The steps:

1. Install or keep your existing SDK.
2. Change the base_url to RelayRouter.
3. Replace the API key.
4. Call the same endpoint (/v1/chat/completions or /v1/messages).

See relayrouter.io/docs for details.

Protocols and endpoints at a glance

RelayRouter supports 2 documented request protocols, each with its own endpoint and compatible SDK.

Both are confirmed by the statement that RelayRouter is compatible with the OpenAI and Anthropic protocols. The documented model families (Claude, GPT-5.5, Gemini 3.5, DeepSeek, GLM, MiniMax, Moonshot) are text oriented in the available facts, with no image endpoint listed.

Billing for failed requests

Failed or errored requests are generally not billed on RelayRouter, according to relayrouter.io. This billing rule applies across the documented protocols and model families, meaning a request that errors out before completing does not normally incur a charge. Because the facts cover only the 2 protocols and the listed model families, any billing specific to image generation (if it were offered) would not be covered here. For accurate, current per-model pricing across the available models, consult the live rates at relayrouter.io/models, which lists each model individually.

FAQ

Can I generate images with gpt-image or DALL-E 3 on RelayRouter?
The provided facts do not document image generation, gpt-image or DALL-E 3. Check relayrouter.io/models for current availability.

Which protocols does RelayRouter support?
Two: OpenAI (/v1/chat/completions) and Anthropic (/v1/messages), per relayrouter.io/models.

Do I pay for failed requests?
No. Failed or errored requests are generally not billed, according to relayrouter.io.