Skip to main content

​
Overview

When you open the model selector in Julius, you’ll notice a set of dollar signs ($, $$, $$$, $$$$) beside each model. These are cost indicators — they show how quickly that model consumes your credits relative to the other options. The more dollar signs a model has, the more credits it uses per action. Model choice is one of the biggest drivers of credit consumption, alongside task complexity, so picking the right tier is the easiest way to control your usage.

​
Cost tiers

Example models are illustrative. Check the model selector in the app for the current lineup and each model’s exact indicator.

​
Why model choice matters

Every action in Julius costs credits, and the total depends on three things:
  • Action type — a simple message costs less than a full analysis or code execution
  • AI model — more capable models use more credits (this is what the $ symbols reflect)
  • Complexity — longer outputs, bigger datasets, and multi-step tasks add up faster
Lighter models like Julius 1.2 Lite and Claude Sonnet 5 consume credits far more slowly than frontier models like Julius 1.2 Max, Claude Opus 5, and GPT-5.6. If you run a simple task on a $$$$ model, you’ll get a similar answer to a $ model — but spend significantly more credits for it.

​
How to stretch your credits further

  • Default to a $ or $$ model for day-to-day work — quick questions, summaries, simple visualizations
  • Reserve $$$$ models for genuinely hard tasks — complex modeling, multi-step analysis, or when a lighter model is struggling
  • Switch mid-conversation — you can change models at any time, so start light and only escalate if the results aren’t good enough
  • Watch your usage — track your credit balance and per-action consumption in real time on the Usage page

​
FAQ

No. They’re a relative indicator of credit consumption, not a direct dollar price. The actual credit cost of an action also depends on what you’re doing and how complex the task is.
Not always. For simple tasks, light and frontier models often produce comparable results. Frontier models earn their cost on complex, multi-step, or high-stakes work.
Yes — use the model selector at any point. A common pattern is to do exploratory work on a $ model, then switch to a $$$$ model for the final, hardest step.