I hit my free tier google api limit immediately, after a single prompt saying hello to my model

[wet bluff]

Is this expected, how can I debug

[solid tangle]

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

[sudden salmon]

A single “hello” shouldn’t normally burn through a Google free tier. When this happens it’s usually one of:

• you’re actually at 0 quota (API not enabled for the project / billing-required tier / quota disabled)
• the API key is tied to a different Google project than you think (and that project’s quota is already used up)
• OpenClaw is retrying/failing over (so one message triggers multiple upstream attempts), and Google is returning 429 / RESOURCE_EXHAUSTED

What to grab so we can tell which it is

1. Version

openclaw --version

2. Which provider/model OpenClaw is actually using + auth status

openclaw models status

3. Logs around the failed request (then send the error block here; redact secrets)

openclaw logs --follow

Run that, send “hello” again, then copy/paste the bit that contains the Google error (status code + message).

What to look for in the error text

• 429 / RESOURCE_EXHAUSTED / “Quota exceeded” → it’s a quota/rate-limit response (we then check whether it’s RPM/RPD or “project has no quota”)
• “API not enabled” / “billing account required” → not actually a “free tier limit”, it’s an enable/billing/quota toggle issue on the Google side
• If you’re using Vertex (google-vertex) vs Gemini API key (google) the quota/billing behavior is totally different — the logs/model status will reveal which path you’re on.

Relevant docs (for the debugging commands + provider/auth expectations)

• Troubleshooting runbook (logs / status ladder): https://docs.openclaw.ai/gateway/troubleshooting
• Provider/auth overview (includes Google env var names + key rotation behavior): https://docs.openclaw.ai/concepts/model-providers

Paste:

• the exact Google error message