Dartmouth Research Computing & Data (RCD) logo
  • Home 
  • HPC 
  • AI 
  • Services 
  • About Us 

  •   Search this site
  •  
  1.   Using Dartmouth Chat
  1. Home
  2. GenAI@Research Computing
  3. Using Dartmouth Chat
  4. What do the tags mean (e.g., FREE, LOCAL, VISION, REASONING)?

What do the tags mean (e.g., FREE, LOCAL, VISION, REASONING)?

When you pull down on the list of models in the upper left-hand corner of the page, you’ll notice a series of “tags” associated with each of them. These tags are meant to help you choose the tool that best meets your needs. These tags include: FREE/$, LOCAL/CLOUD, REASONING, HYBRID REASONING, and VISION.

Please see below for a description of each set of tags.

Choosing a Model

The tags:

  • FREE/$$$. This tag helps you distinguish the relative cost of using third-party models. Models marked as “FREE” are local to Dartmouth and can be used without affecting your daily quota of tokens. The relative cost of using cloud-based models can be seen by the number of $ signs displayed - i.e., the more $, the more expensive they are to use. If you are considering a specific project, you can use the Dartmouth Tokenizer tool to understand what your token use might look like when using specific models.

  • LOCAL/CLOUD. Models marked with this tag are local to Dartmouth. This not only affects the relative cost of using the model (i.e., these models are available for “FREE”), but also the relative security of the data shared. Models marked as “LOCAL” are hosted on Dartmouth computers. Data sent to the local models (like chat prompts and/or uploaded images and files) never leave Dartmouth’s systems nor are they shared within the College for any purpose.

  • VISION. This tag highlights models with specialized capability. VISION models are able to understand and process information from both text and images, and can review and process information contained in images.

  • Reasoning. Reasoning allows the model to draw connections between different data, make judgments, and weigh the benefits of potential actions. Of note, tokens are expended during this process even if the “thinking” isn’t displayed in the user interface. In other words, reasoning tokens are billed tokens, so using reasoning consumes credits more quickly. There are two options available in Dartmouth Chat: REASONING and HYBRID REASONING.

    • REASONING. Reasoning models use this feature by default. OpenAI has developed a series of reasoning models, including o4 Mini which is available through Dartmouth Chat.

    • HYBRID REASONING. These models have the ability to combine structured knowlege with data-driven insights, but by default they respond directly like a regular model.

    You can control the degree of reasoning used by adjusting the Reasoning Effort parameter from the Chat Controls.

Didn’t find what you needed? Please reach out to research.computing@dartmouth.edu.

 How do I navigate the Dartmouth Chat webpage?
How do I choose the right model(s)? 
Copyright © 2025 Dartmouth Research Computing & Data | Powered by Hinode.
Dartmouth Research Computing & Data (RCD)
Code copied to clipboard