Adapted community discussion: Paraphrased and condensed from an anonymized public OpenClaw Discord support conversation from September 2026. Technical guidance was reviewed for OpenClaw 2026.9.3.
Discord messages reach my Gateway, but the agent fails before replying with No route-compatible authentication source is configured for openai. The failed message stays queued for retry.
That is model-auth routing, not Discord ingress. The message arrived, but the selected OpenAI route has no compatible credential.
Run openclaw models status and decide which credential type you intend to use: ChatGPT/Codex subscription sign-in or an OpenAI Platform API key. Keep all tokens and redirect credentials private.
I want to use my ChatGPT/Codex subscription, not a Platform API key.
Use the provider login on the Gateway host:
openclaw models auth login --provider openai
openclaw models auth list --provider openai
openclaw models status
Complete the browser sign-in and do not paste its callback or token into a chat.
OAuth is now listed, but the configured model still says it requires API-key authentication. It looks like the sign-in succeeded while the route is using a different auth method.
Yes. That is a route conflict, not a failed OAuth login. Inspect the selected default and the auth inventory:
openclaw config get agents.defaults.model --json
openclaw models auth list --provider openai
openclaw models status --json
An upgraded configuration can retain a custom OpenAI provider entry or an older model reference that resolves through the API-key path. Run openclaw doctor --fix, then use the model configuration flow to select the subscription-compatible route. Do not add an API key merely to silence the warning.
Current reference: OpenAI provider.