Changing model from hermes-agent to another name in the Hermes Agent API has no effect, not because your request was not sent, but because the official documentation has explained that this field is currently mainly front-end compatible, and the actual model used is still the server-side configuration. In other words, Hermes will not really cut to another LLM based on what model name is passed on the frontend.
Where is the real control of the model
The actual model is currently determined by Hermes Agent's own provider and configuration, not the API request body. If you want to change the model, do these things first:
- Change Hermes' provider/model configuration.
- Switch with
/modelin a CLI or gateway session. - If you want to use different models for different scenarios, it is more stable to divide profiles or instances.
So why keep the model field
Because many OpenAI-compatible frontends must see a model to work, such as Open WebUI, LobeChat, and various SDKs require this field. Hermes Agent accepted it to be compatible with these front-ends, not to let it take over the underlying mold selection logic.
In one sentence: the model in the front end is more like an access card than a steering wheel. You need to change the model to change the Hermes Agent's own configuration, not just the request body.
Official open source address: https://github.com/NousResearch/hermes-agent; Official document entry: https://hermes-agent.nousresearch.com/.
