Model Selection
How Innovatic chooses the right AI model for each task — and how to override it.
Every agent conversation uses a specific AI model. Innovatic can choose for you automatically, or you can set it yourself before starting.
Auto mode (recommended)
By default, every new conversation starts in Auto mode. Innovatic analyzes your first message and selects the model that best matches the task:
- Simple, well-scoped tasks — get a fast, lightweight model for quick turnaround
- Most development work — get a balanced model for the right mix of speed and quality
- Complex reasoning, architecture, or nuanced tasks — get the most capable model with extended thinking enabled
After you send your first message, Auto is replaced by the chosen model name, and you can see exactly what was selected.
Manual selection
If you prefer to control model selection yourself:
- Click the model selector before sending your first message
- Choose from the available models in the dropdown
- Toggle extended thinking on or off using the thinking toggle that appears
Two keyboard shortcuts let you cycle through options quickly:
| Shortcut | Action |
|---|---|
Alt+P | Cycle models: Auto → Opus → Sonnet → Haiku → Auto |
Alt+T | Cycle thinking levels (only available in manual mode) |
Settings lock
Once you send your first message in a conversation, the model, thinking level, and work mode are locked for the rest of that session. The selectors remain visible so you can see the active configuration, but they can't be changed.
To use a different model or mode, start a new conversation.
Auto mode is the right choice for most users. It eliminates the need to think about model selection on a per-task basis and consistently delivers the right trade-off between speed and quality.