Skip to Main Content
- Copilot is the only LLM currently approved by Gonzaga ITS’s AI Guidelines for use with Gonzaga data, if the user logs in to the enterprise version of Copilot with their Gonzaga Single Sign-On credentials.
- Look for the green check in the upper right showing that enterprise data protection is enabled.
- Copilot uses a version of OpenAI’s ChatGPT 4, with custom tuning by Microsoft.
- As of this writing, the downloadable desktop app does not work with the enterprise version of Copilot, so GU users should use the web app at m365copilot.com.
- When retrieving information from the internet, Copilot generates a text search in Microsoft’s Bing search engine
- Has two modes of prompting "deeper thinking" in which the AI takes more time and steps before providing a response
- Can generate images
- Available in a free tier using an older and less capable model, in a paid tier which allows access to newer and more capable models as well as multimodal capability (the ability to handle text, images, code, audio, and uploaded files).
- Available both on the web and in a dedicated app for desktop or phones
- Can generate images
- Has a "deep research" model which is available to free users with usage limits, and to paying users without limits
- Available in a free tier using an older and less capable model, in a paid tier which allows access to newer and more capable models.
- The free tier can use a limited amount of multimodal input (uploaded slides, PDFs, etc) (10 as of this writing), with the only limit for paid users being the context window
- Tends to have a larger context window (the amount of user text and data it can work with) than ChatGPT
- Available via web and phone app. There is no native desktop app for Mac or Windows.
- Can generate images and (for paid tiers) video
- Uses Google for search, of course
- Has a "deep research" model which is available to free users with usage limits, and to paying users without limits
- As of this writing, Gemini is often described as being a little less capable of "creative" output than ChatGPT, but a little better at tasks like data analysis, research, and coding
- Created by Anthropic, an AI company started in 2021. In addition to creating AI tools, Anthropic does a lot of interesting research about AI.
- Available in a free tier using an older and less capable model, in a paid tier which allows access to newer and more capable models and greater usage limits.
- Can analyze uploaded charts and images, but not audio or video
- Does not generate images or video
- Can create "Artifacts" - interactive content, documents, or code that the user can see and modify during the conversation.
- Has a "deep research" mode
- Has Mac, Windows, and Phone apps
- From Meta, i.e. Facebook
- Versions of Llama can be downloaded and run locally
- Can generate images
- No "deep research" mode
- From "X," i.e. Twitter, Grok's main claim to fame is that it has fewer guardrails than other major models. This can lead to . . . interesting behaviors.
