Understands intent automatically
Classifies each prompt (writing, coding, research, analysis, creative work and more) and picks a model suited to that kind of task.
feature
Auto Mode weighs speed and depth so you get a suitable answer without juggling providers, and you can take manual control at any time.
Classifies each prompt (writing, coding, research, analysis, creative work and more) and picks a model suited to that kind of task.
Routes lightweight prompts to fast, efficient models and reserves the deeper reasoning models for work that genuinely needs them.
Uses extended reasoning on models that support it when a task benefits, and sticks to low-latency responses when speed matters more.
Auto Mode is a default, not a cage. Pick any model from the picker at any point and the conversation carries on with full context.
Auto Mode handles the heavy lifting, but the moment you need a specific model, manual controls are one click away.
Choose a specific model from the picker whenever you want. Switching mid-conversation keeps your history and context intact.
Because every model lives in the same interface, you can draft with one model and refine with another without starting over.
A message routed by Auto Mode counts exactly the same as any other message on your plan. Routing never adds a surcharge.
Prefer manual control? Pin your favourite model in Chat or browse the Model Directory.
Auto Mode classifies each prompt by task type, then selects a model suited to that category, balancing response speed against reasoning depth. Simple questions go to fast models; complex work goes to deeper ones.
Yes. You can switch to any model in the catalogue at any time from the model picker, and the conversation continues with full context.
Yes. When new models are added to the registry, routing is updated so Auto Mode can take advantage of them.
No. A routed message counts as one message on your plan, the same as a message sent to a model you picked yourself.