Skip to main content
The chat toolbar gives you direct control over which AI model Hermes uses and which profile it runs under. The Model dropdown and Profile selector live in the toolbar and update live — no need to edit configuration files or restart VS Code manually.

​
Choosing a model

The model dropdown auto-populates from Hermes when a session connects. Models are fetched from the running Hermes process and grouped by provider. If the live list is unavailable, the extension falls back to a static list you configure in settings. What the list shows:
  • Models are grouped by provider (for example, Anthropic, OpenAI, and any custom gateway you have configured).
  • Each model row displays the output cost per 1M tokens so you can compare pricing at a glance.
  • The active model is highlighted and the containing group is listed first.
  • Long lists scroll inside the panel without clipping.
To switch models, click the Model button in the toolbar, then click the model you want. The session restarts immediately with the new model active.
Switching models resets the agent’s in-memory context for the current session. The chat history is preserved locally and flagged with a local-history banner, but the agent does not retain it. Use context attachment after the reset to carry prior messages into the new session.

​
Hermes profiles

Profiles are auto-discovered by running hermes profile list when the extension starts. Any profiles defined in your Hermes configuration appear in the Profile dropdown in the toolbar — no manual ID juggling required. Click the Profile selector to open the list and click a profile to switch to it. Like a model switch, changing the profile restarts the agent session.
Set the hermes.profile setting to make a profile the default every time the extension starts. If you work across multiple Hermes configurations, use hermes.agents to define named agent configurations (each with its own path, profile, and working directory) and switch between them from the toolbar.

​
Fallback model list

If Hermes is not running or does not expose models, the extension falls back to a static list you define in your VS Code settings. Configure it by adding the following to your settings.json: 
"hermes.models": [
  { "id": "claude-sonnet", "name": "Claude Sonnet" },
  { "id": "gpt-4o", "name": "GPT-4o" }
],
"hermes.defaultModel": "claude-sonnet"
The hermes.defaultModel value must match an id from the hermes.models array. The fallback list is only used when Hermes does not return a model catalog; it is replaced by the live list as soon as a connection is established.