Model Switching
Model Switching lets you choose which configured model your Hermes Agent uses for its next work. Switch when a task needs a different balance of speed, capability, cost, or provider availability—without creating a new Agent.
This page covers
- Preparing more than one usable model
- Selecting the active model for an Agent
- Verifying the change in Chat
- Troubleshooting a model that does not appear or cannot respond
Good to know: An Agent uses one active model at a time. Switching the active model does not erase the Agent's memory, Cloud Storage files, skills, or automations.
Before you start
- Open the Agent you want to update.
- Make sure the target model has already been configured and is available to your account. If not, follow Configure Models first.
- Keep a simple test prompt ready so you can confirm the new model responds as expected.
Switch the active model
- Open Agents and select the Agent.
- Open the Models section.
- Choose the configured model you want the Agent to use.
- Save or apply the change, then wait for the Agent to finish updating its active configuration.
If you are choosing a model for the first time, use Configure Models to add a supported provider or custom model before returning here.
Verify the change
Open Chat and send a short, low-risk prompt, for example:
Reply with the name of the current model if it is available, then summarize this sentence in one line: Model switching is ready.Confirm that the Agent responds normally and that the model shown in the Agent console is the one you selected. If the first reply takes a little longer while the Agent updates, wait briefly and retry once before making another change.
Choose the right time to switch
| Situation | What to consider |
|---|---|
| You need a faster first draft | Choose a configured model that fits short, routine tasks. |
| A task needs stronger reasoning or code help | Choose a model you have configured for that kind of work. |
| A provider is unavailable or rate-limited | Switch to another configured model, then test it in Chat. |
| You are comparing output quality | Run the same small prompt after each switch and compare the result. |
Model Switching changes the engine used for future responses. It does not move your conversation history, rewrite files in Cloud Storage, or change your Skills and Automations.
If a switch does not work
- Return to Configure Models and confirm the target model, credentials, and provider settings are complete.
- Confirm that the model is available to your account and has sufficient provider quota or credits.
- Select a known-working model and test it in Chat.
- If the Agent still cannot respond, open Diagnostics and review recent logs before resetting its model configuration.
FAQ
Why is the model I want not listed?
Only configured models can be selected. Add or complete the model in Configure Models, then return to the Models section.
Does switching models clear my Agent's memory?
No. The active model changes, while the Agent's memory and saved setup remain in place. You can review memory separately in Memory.
Can I switch back if I do not like the result?
Yes. Select another configured model, apply the change, and test it with the same prompt. Keeping the prompt consistent gives you a more useful comparison.
Next steps
- Set up additional providers in Configure Models.
- Use Chat to compare model responses on a real task.
- Use Diagnostics if a configured model repeatedly fails to start.