The Hermes Agent cron task did not automatically switch to fallback after failing, which is not necessarily a misconfiguration. The official Fallback Providers documentation states that cron jobs do not support automatic fallbacks like the main session, and they will run according to the provider configured at the time of execution. If you want to use different models, you need to specify the provider and model in the cron job itself.
Correct understanding
fallback_model in the CLI session can be switched at master model 429, 500, 401, 404; However, cron is a timed task and does not inherit this session-level bottom. You want to configure cron as a standalone job.
How to reduce the failure rate
- Explicitly specify a stable provider for the cron task.
- Break down key tasks to avoid running for too long at a timed task.
- If it is often 429, give the provider a key pool or change to a more stable model.
- After the failure, use the logs to confirm whether it is an authentication, current throttling, or model capability issue.
In a word: cron does not fall back is the current mechanism, and the solution is to specify a provider/model for the task itself.
Official open source address: https://github.com/NousResearch/hermes-agent; Official document entrance: https://hermes-agent.nousresearch.com/.
