Skip to main content
Reasoning has Julius think through a problem before it answers. With reasoning on, Julius plans the task, works through it step by step, and checks its own work as it goes. It helps most on complex, multi-step questions. Turn it on with the Reasoning button in the chat composer.

​
Choosing a reasoning level

If your selected model supports reasoning, you can set how hard Julius thinks. Levels run from low to high, and some models add an extra-high setting. A higher level means Julius spends more time reasoning before it answers, which pays off on harder problems. The available levels depend on the model. If a model doesn’t support reasoning, the button stays off.
Adjusting the reasoning level is available on paid plans.
Higher reasoning levels use more compute, so they spend more credits. Turn it up for genuinely hard tasks, and keep it low for quick questions.

​
When to use it

Reasoning helps when you want Julius to handle a complex task in a single prompt, like cleaning data, running an analysis, and summarizing the findings together. For the most control, you can still break a big problem into smaller prompts and iterate: clean the data, then explore the drivers, then summarize. A step-by-step approach and a higher reasoning level work well together. Turn up reasoning when:
  • The task has several steps, or each step depends on the last
  • You’re comparing scenarios or weighing trade-offs
  • You want a structured plan, checklist, or analysis in one go