Hermes "quota exhausted" on Gemini when quota remains
Hermes can't distinguish short-term RPM/TPM limits from true daily quota exhaustion, so it reports 'quota exhausted' and blocks until the reset timer even when daily quota remains. The google-gemini-cli OAuth path is especially tight. Wait out the timer if genuinely limited, or switch config.yaml to an API-key path using an AI Studio key and lower concurrency.
Likely cause
Hermes can't tell short-term RPM/TPM limits from true daily quota exhaustion, so it reports 'quota exhausted' and blocks until the reset timer even when daily quota is left. The google-gemini-cli OAuth path has especially tight concurrency.
The fix
- 1 Wait out the reported reset timer if you are genuinely rate-limited.
- 2 Switch the model line in config.yaml from the google-gemini-cli OAuth path to an API-key path using an AI Studio key.
- 3 Lower concurrency so you stay under the per-minute limits.
Hit a different error?
Paste any agent error and get the cause and fix in seconds.
Frequently asked questions
I still have daily quota left - why does it say exhausted?
Hermes can't tell a short per-minute (RPM/TPM) limit from a full daily-quota wall, so a brief burst reads as 'quota exhausted' and it blocks until the reset timer. Lowering concurrency usually clears it.
Is the OAuth path or the API key better?
The google-gemini-cli OAuth path has tight concurrency limits. Switching config.yaml to an API-key path with an AI Studio key generally gives more headroom for agent workloads.
How do I avoid hitting the per-minute limit?
Reduce parallel requests and subagent concurrency so you stay under the RPM/TPM ceiling. That stops the transient limit from being misreported as a daily quota exhaustion.
Stop firefighting agent errors
Decoding errors one at a time is the manual version of what BetterClaw automates. Run your OpenClaw agents hosted with managed models, retries and config validation built in.
$19/month per agent · BYOK · 7-day money-back guarantee