Forum

Hi Jamale,

Honest upfront: OpenClaw OpenRouter setup is a product-level config question that the OpenClaw community will answer faster and more accurately than us — we can offer general direction but not authoritative "this exact env var on this exact OpenClaw version" answers. The right place for definitive help is OpenClaw's GitHub Discussions / Discord.

That said, here's what we typically see for OpenRouter + OpenClaw setups:

Common env vars (check what OpenClaw's docs say, since names vary by version):

OPENROUTER_API_KEY=sk-or-v1-...
OPENROUTER_BASE_URL=https://openrouter.ai/api/v1

Or sometimes OpenClaw uses OPENAI_API_KEY + OPENAI_API_BASE (treating OpenRouter as an OpenAI-compatible endpoint):

OPENAI_API_KEY=sk-or-v1-...
OPENAI_API_BASE=https://openrouter.ai/api/v1

After setting env vars in Service Settings → Variables, restart the service for changes to take effect.

For "only 2 models showing":

OpenClaw's model picker usually populates from one of:

• A static list in OpenClaw's config (need to add OpenRouter model IDs there)
• Dynamic fetch from the API endpoint (OpenRouter has /api/v1/models endpoint)

If only 2 models show after configuring OpenRouter, OpenClaw is probably not fetching the dynamic list — needs explicit model entries added to its config file. Check OpenClaw's config (usually config.yaml or similar) for a models: section.

For "errors when chatting":

Open OpenClaw's Service Logs in Zeabur dashboard:

• Dashboard → your OpenClaw service → Logs tab
• Trigger a chat that errors out
• Look for the error line right after your message
• It'll tell you exactly what's failing (auth, network, model name, etc.)

Paste the error log here if you want us to look — we can help interpret what the error means, even if "how to fix in OpenClaw's specific config" is a question for them.

For "internal Zeabur models give nothing":

If OpenClaw is configured to use "Zeabur AI Hub" as an inference backend, two common issues:

1. AI Hub credits depleted — Dashboard → AI Hub → Balance Management. If $0, top up
2. AI Hub API key not set in OpenClaw — needs ZEABUR_AIHUB_API_KEY or similar env var

To get unblocked fastest, please reply with:

1. Service Logs content when you trigger a failing chat (paste the error text)
2. Screenshot of your Variables tab showing the env vars you've set (mask the actual API key)
3. OpenClaw version you're running (in Service Settings → Source → image tag)

With that + checking OpenClaw's own docs for the exact env var names for your version, we can probably pin it down. For "official authoritative OpenRouter integration setup", also post in OpenClaw's GitHub Discussions — they'll have copy-pasteable working configs.

Sorry I can't be more specific without seeing your setup — Zeabur platform's role here is just "host the container", and the OpenClaw + OpenRouter integration logic happens entirely inside the container.

k thx i will go down your list of fixes, had gotten a headache the other day, from this....i eventually got things running on my local machine, "local ai" mac mini...im just trying to get something on vps. for learning puposes...i hav other instances on other vps running no problem as well, i guess i just like troubleshooting....lol