Configure your Companion model
Control the underlying technology that powers your Companion. You can select your preferred AI provider and specific model to tailor the quality, speed, and reasoning capabilities of your assistant.
The Models tab controls which AI provider and model Companion uses for your new threads in the current workspace.
To access these settings, go to User preferences > Models. The underlying URL may still include intelligence, but the visible tab is Models.
Note: These settings are workspace-specific. Changing your model in one workspace does not change it in another.
Select an AI provider
Harmony supports a broad catalog of providers, including OpenAI, Anthropic, Google, Mistral, Groq, and OpenRouter-powered models. The exact list you see depends on what your workspace admin has enabled.
Choose a provider tile first. Harmony then shows the models available from that provider.
Choose a model
Use the Model dropdown to choose the model Companion should use for new threads. If your saved model is no longer available, Harmony falls back to the provider or workspace default until you choose another one.
Pick based on the type of work:
- Use faster, balanced models for quick summaries, drafting, and day-to-day questions.
- Use stronger reasoning models for complex analysis, long documents, strategy, or code.
- Use smaller or lower-cost models for high-volume repetitive work.
- Use models with vision or tool support when your task requires those capabilities.
If a model you want is missing, ask a workspace admin to enable it under Workspace Settings > Intelligence.
Changing models in a thread
You can choose a model for a new thread before sending the first message. Once a thread has messages, its model is locked. Start a new thread to use a different model.
Reset to defaults
The Reset to Defaults action removes your custom model choice and returns to the workspace default. It may also reset other Companion preference fields, so review your Companion tab after using it.