AI Models and Tools

  • For workshop prompts, click here.
  • For workshop slides (with 60+ pages of citations): click here

Clicking the links below will open the listed AI tools in new tabs in your browser. The list begins with LLM models (proprietary, open-source, regional and reasoning) and then a range of API tools (followed by agents and browser extensions)

PROPRIETARY FRONTIER MODELS: Here are some AI models you should know. They are from different companies, using different different neural networks and with different personalities and abilities. The paid versions are often substantially better and smarter. An easy way to compare models is to use ChatHub (which makes it easy by putting responses side by side) or Poe, which is probably the best way to get started: for $10/month you can try ALL of the best tools (including the paid voice, image and video tools) in one place. If you want to just pick one, for most people and uses, paid ChatGPT is the place to start–its Code Interpreter is best for statistical analysis. Ethan Mollick has written an excellent summary (Jan 26, 2025) of the differences and how to pick one. 

THE BIG THREE 

  • Claude.ai Lots of tokens for writing context and designed with a constitution, to do no harm. Sonnet 3.5 is GREAT and often free. There is also a smaller Haiku and a larger Opus, but Sonnet 3.5 is by far the best version to use. Claude can run code, can analyze images and charts in pdfs, but cannot actively search the web at the moment.
  • ChatGPT You need ChatGPT 4o (omni) which is faster but similar to 4.0. You need to see 4o in the upper left (under the ChatGPT menu OR see 4o in the address. GPT 4 can search the web (for ideas and not just words) is multimodal and can now watch live video and interact with voice. Watch a demo here. Copilot is really another version of ChatGPT (Microsoft owns half of OpenAI) that integrates with Microsoft projects–so it is better with Excel and Ppt, but none are great with that yet. If your organization gives you access to this in MS Office you also are FERPA and HIPPA secure. In CoPilot you can try Think Deeper for (free access to the open OpenAI o1 reasoning model).
  • Gemini (formerly Google Bard) is the third powerhouse model. For an adult you need Gemini 2.0 Flash which can also search the web (2.0 opens the door to other features like DeepResearch, LearnLM and graphing (see below). Project Astra and Project Mariner (which can control your browser). Note that after you get a response, you can click on the three dots to access the “Double-Check Response” which will provide verification for elements it things needs it. Google has a deal with the Associated Press to get real-time news updates.) Also try the new multimodal Gemini Studio (using Gemini Flash 2.0 or other experimental models like LearnLM) for free. Gemini is also multimodal and can watch live video and interact with voice

OTHER MODELS

  • Sonus is a new “set”family” of proprietary models (Pro, Air, Mini and Pro with Reasoning, see below) that is already competitive with the very best existing models. For the moment, this is the best (only) way you can try the new reasoning models for free: here.
  • WolframAlpha combines the computational powers of Wolfram|Alpha with ChatGPT.
  • Pi is focused on dialogue and role-playing and has a good voice interface.
  • You.com is set up to be a search engine competitor to Google but with more privacy and easier customization (so it now faces competition from Gemini, but also ChatGPT Search). 
  • Grok 3 (released February 17, 2025) is a new family of models (including a mini version) that might at last be in a similar class to 4o and Gemini. It can analyze images and has real-time access to the social network X. 
  • Amazon Nova are both good Class 2 models (so on a par with GPT 4).
  • Poe and ChatHub are consolidators that provides access to multiple AI through one interface.

OPEN SOURCE MODELS: There are now open source models that are just as good as the best proprietary frontier models, and even better in some specialized areas. You can download most of these models from Github, Azure (Microsoft) or HuggingFace. You can then fine-tune and run them on your laptop, which deals with most privacy issues but also transfers the security risk to you.

  • Chinese DeepSeek is strong with text and can also search the web. You can try the R1 model (which “reasons” like o1) here. It is a cheaper API option and it was built for a fraction of the price/chips/energy of the big models through the clever use of Multi-head Latent Attention (MLA) that combines even more values into tokens (the simple version is tokens that read phrases, so less precision but turns out it was not needed and not all tokens are active all the time for a huge energy, cost and time savings). Here is a great non-technical summary of how DeepSeek is important or you could read this tech paper.
  • Meta AI is the chat version of Llama3.2 which also a Class 2 model. It does not require a login.
  • Kimi is an excellent free multimodal open source Chinese AI that has a particularly large context window (good for long papers, prompts, and conversations – you can upload 50 files 100MB EACH), does very well in math and coding (beating GPT-4o and Claude Sonnet 3.5 on Codeforces), searches the web, can analyze charts and also has reasoning.
  • Qwen 2.5 from Alibaba comes in a range of levels (select at the upper left in Qwen Chat). Qwen2.5 beats that beats GPT-4o, Claude-3.5 Sonnet, and DeepSeek-V3 while Qwen2.5-1M can handle up to one million tokens. Qwen2-Math scores high (maybe the highest for pure LLMs?) at math.
  • Mistral (available both as an API and Le Chat) is an open source LLM from France that real time internet search (with press wire access for news) and is very fast. It also creates great images using Black Forest Labs Flux Ultra. 
  • Huggingface is a chatbot running on Llama. Start here to get a sense of what open source can do. No login is required.
  • Llama 3 from Meta, with 70B parameters is very close in performance to the best paid models.
  • Falcon (Mamba 7B) is an open source LLM from the UAE uses new “state space” architecture (SSLM)instead of the transformer architecture.

REGIONAL and CULTURALLY-SPECIFIC MODELS: Since people and cultures think differently, we are starting to see LLMs that are trained on culturally specific data sets. Note that if you want a culturally specific answer, you can and should still try this with the frontier models (try asking Claude to response as a Black professor and compare the response to Latimer).

  • Latimer (named after African-American engineer Lewis Latimer) aims to better represent diverse communities by adding further training from (verified and licensed) books, oral histories and sources from Black and Brown communities.(Latimer is a fine-tuned version of LLAMA.)
  • Mistral Saba is a 24B parameter version of the French Mistral model trained on curated datasets from across the Middle East and South Asia. It supports Arabic and many Indian-origin languages like Tamil.  
  • Fanar is a “culturally and regionally aware” Arabic LLM fluent in Arabic dialects from the Qatar Ministry of Communications and Information Technology (MCIT) and the Qatar Computing Research Institute of Hamad Bin Khalifa University (HBKU).
  • LatamGPT from Chile’s National Center for Artificial Intelligence (Cenia) is an open source model trained on “characteristic data from Latin America” and is due to launch in summer of 2025.
  • There is also a group of behavioral scientists exploring the training of LLMs on historic texts Viking, Latin or Medieval Arabic etc.) 

REASONING MODELS: The next big thing are models that think through problems before answering. You need to use them a little differently: give it something hard to do and note (or ask) how it describes its reasoning. Look at this example. The progress here has been rapid and substantial (read this report about the new o3 from Dec 2024), but in mid-January 2025 they were all behind paywalls. You can try these for free:

  • ChatGPT (o3-mini) is activated when you click the “Reason” button in the chat box. You can also access OpenAi’s o1 model through CoPilot by clicking the Think Deeper button.
  • Gemini Flash 2 Thinking Mode can be access in Google’s AI Studio. 
  • Sonus Pro with Reasoning (a new model from a start-up!) Select the Pro version with Reasoning turned on) but Open sources models appear to be keeping up. 
  • DeepSeek’s v3 R1 is also open source which means you can download and build with it. You can try DeepSeekR1 on an American server using Perplexity (Pro Search). Here is a great non-technical summary of how DeepSeek works that includes a good summary of how reasoning models work.
  • Qwen 2.5 Max is another Chinese open source reasoning model. Select the 2.5Max model in the upper left when you are in Qwen chat. It is also available in HuggingFace.
  • Kimi also has reasoning, but I’ve not tried it yet.

EpochAI is an important independent organization that is keeping track of these models, how they compare and where we might be going. They maintain a great dashboardcomparing capabilities of the best models (against their own benchmarks) and also this larger data set of virtual all models. They produce excellent reports about trends including a recent prediction that AI will continue to improve rapidly.

APIs = APPLICATION PROGRAM INTERFACE. This is a huge category and most of the new products you see are here using an API to interact with one of the frontier models for you. You can often replicate the results with longer and careful prompting, but these are very useful shortcuts. 

APIs for INFORMATION, LITERATURE SEARCH & RESEARCH

  • Perplexity.ai is an AI-powered chatbot search engine. It answers your questions with the sources cited using multiple frontier models. A competitor (still experimental) is Google’s Learn About, but for getting the gist of a topic, these are useful. Perplexity also has an Internal Knowledge Search that will search your files for info. 
  • Consensus.app and Elicit.com are academic research tools that limit data search to the 200M published papers in Semantic Scholar and use AI (ChatGPT) to allow you to filter by claims, methodology, sample size and more. Consensus includes a “consensus meter” that provides an estimate of the consensus in the published literature. Here is the result when asking “do brain games work?” Elicit has a filter that allows you to search by the quality of journal.
  • Storm (short for brainstorm) is a new research tool from Stanford that creates a Wikipedia-like report on the topic of your choice. It looks at more than just Semantic Scholar publications. It will write/summarize from different perspectives (ex. sociologist vs political scientist) and tell you what sources it used. Compare the results and format with what you get from Consensus. 
  • Here is a comparison of Consensus, Elicit, Storm and Perplexity answering the question “do polls predict elections?”
  • Researchrabbit and Litmaps are both more visual, showing you a network of articles that relate to a topic. It is a bit like Spotify for research papers (you liked that, you should know about this).
  • SciSpace also has similar functions but with a broader suite of tools, like a paraphraser that rewrites or helps explain passages (something you can also find in ExplainPaper.) All four of these are essential lit review tools. 
  • Scite extracts citations and uses AI to analyze if they are cited with support or contradiction in other papers. Upload a pdf or your citations and find out quickly about the impact of the work.
  • Semantic Scholar a free AI search tool with a pdf reader.
  • DeepResearch (available only with Gemini Advanced) creates a multi-step research plan. If you approve, it searches the web, analyzing relevant information. It repeats this process multiple times to generate a report of the key findings with links (which you can export into a Google Doc). The difference from the other research tools is this is agenic: it searches the web and repeats the process. 

APIs for ANALYSIS & WORKING with DATA. Here is a subcategory of tools that allow you to control the data set or knowledge base:

  • NotebookLM is Googles version of a research assistant but it works only on the documents (up to 50) you upload (up to 500,000 words EACH). Use it as a virtual committee assistant: link or upload all of your institution policy documents plus the committee charge and then faculty can interrogate s needed. Try uploading a book and asking for a study guide or an interactive podcast. Here is an AI-created podcast about the first part of my Teaching Change book.
  • Mem has similar features that allow you to “chat with your data.” Perplexity calls its version of this Spaces.
  • There are lots of “chat with your data” sites that pitch themselves as “research” tools, but are really aimed at students. Sites like Scholarcy, Mindgrasp (“Learn 10x faster!”) and CoralAI, (“read documents faster”) focus on summaries and audio transcriptions (like lectures you did not attend). 
  • The tech behind the podcast feature in NotebookLM is Illuminate, which has more features: you can change the voices, accents or styles and turn any content or webpage into a dialogue. A competitor is GenFM from ElevenLabs.
  • Nomic Atlas îs best for analysis of huge unstructured data sets and does a range of visualizations. Julius also allows you to do computations and visualizations with your data and also writes reports, finds insights and does analysis.
  • Ailyze is focused on qualitative data and insights and also has an AI-driven interviewer to help with collecting data.
  • Napkin creates infographics and visualizations from your text. You don’t have to prompt–just drop in the items you want to visualize and you get lots of options that you can then alter with easy tools. Infography also makes infographics from text.

APIs for WRITING, GRADING, TUTORS and MORE

  • WRITING: Grammarly, and Quillbot are already known to students (before they were enhanced with AI) and (along with the newer Caktus) blur the line of improving with cheating, offering to write paragraphs, solve problems, answer questions and check if your content can bypass AI detectors. Copy.ai and Jasper and many many others focus on specific types of writing or business uses.
  • GRADING: Your choice here is either to use one of the best (smartest) frontier models (see above) but understand that is is naive and you will need to provide lots of very specific instructions (it needs a recipe) or you can use one of these already fine-tuned models (that already knows how to cook) that often use and older, cheaper and less smart (GPT 3.5) to do the work. There are already over 70 AI grading tools available.
    • CoGrader does general grading and feedback and integrates with Google classroom. Try the 2.0 version here: https://v2.cograder.com/
    • TimelyGrader can import assignments from your LMS and export grades back to it.
    • AI For Teachers: Free for ChatGPT Plus users
    • Gradescope: Also from Turnitin.
    • Kangaroos AI: Customizable rubrics and bulk uploads.
    • EssayGrader: Free option (with limited rubric customization) and allows bulk uploading.
    • Smodin: Limited free version but includes language translation.
    • GradeCam: Instant feedback that integrates with some LMS.
    • SnapGrader: Includes a scanning feature. 
  • ChatBot Assignments with Grading: Mostly K-12 at the moment, but look for new platforms like Parlay and Mizou that create specialized chatbots for problems or situations and then grade student interactions–all within a protected environment.
  • NOTE-TAKING: Microsoft OneNote, Otter.ai, Fireflies, and Zoom Companion all do more than just transcribe notes, organize and summarize (often across multiple documents/meeting ). They can analyze who is talking (or interrupting) the most and some even the emotions of participants. Find the latest list of “best” on your platform.
  • STUDY ASSISTANTS: Both Nurovant, Turbolearn.ai and GradeMaxx (and MANY more) can turn lecture notes, a pdf, a phone recording or a YouTube video into an outline, practice tests, flashcards, mind maps, quizzes or notes. Learn About is hoping to be a big player in this space.
  • NotebookLM also does this: try uploading notes, pdf (or up to 50 documents of content) and it will create a summary, sample questions, study guide or even a podcast: here is an AI-created podcast about the first part of my Teaching Change book. (What about an assignment that asks students what information is missing?)
  • Storm is a new research tool from Stanford that creates a Wikipedia-like report on the topic of your choice. 
  • There are a growing number of more visual ways to explore new topics and all of these create concept maps (and each does more):  Mapify, Albus, and Heuristica(which also has templates for pro/con boxes, controversial points, flash cards and more)
  • FINE-TUNED BOTS & TUTORS: Each of the big platforms also has a way to build and then distribute your own fine-tuned applications: GPTs (from OpenAI), Assistants(from HuggingFace), Bots (from Poe). Faculty developed writing tutors, for example, include one from Mark Marino, AI Tutor Pro from a group of Canadian faculty and MyEssayFeedback in beta from Eric Kean. 
  • SchoolAI, MagicSchool and Khanmigo all provide tools to help with specific tasks that are free, FERPA compliant and secure. This includes creating specialized tutors. In SchoolAI go to Spaces and then Create. You can simply prompt it (Help students master content X by providing an overview and asking questions etc) or you can upload documents and set a standard for mastery. Importantly, SchoolAi also has a backend that tells you have students have engaged and what they might still be confused about. Here is a great example (solving Linear Equations in One Variable from Rebecca Tyler at Great Falls College MSU).
  • Snorkl provides feedback to student on their verbal or visual thinking.
  • How to Build Your Own Customized Chatbot (free chapter from Levy and Albertos (2024 Teaching Effectively with ChatGPT.
  • How to use Speaker Progress in Microsoft Teams to get feedback on your/student presentations

VOICE AND DIALOGUE: This exploded in May 2024.

  • Pi is focused on dialogue and role-playing–click the speaker icon in the upper right to have a conversation.
  • Hume.ai is in beta but claims to be even more emotionally intelligent than Pi.
  • GPT 4o in May 2024 was a real game-changer and it still has the best Live Mode in its Advanced Voice Mode. Claude and Gemini can also talk to you now. Ask it to create a customized bedtime story. Watch some of the videos to get a sense of how seeing and hearing (and interrupting) might be leveraged as a tutor or for translation.
  • The ending behind the podcast feature in NotebookLM is ElevenLabs, which offers more customization options but also does other text to audio conversions. Cartesiais a competitor is this evolving text to audio world. 
  • Kokoro is an open-source model that allows you to create natural sounding speech from text. 
  • See Communication and Relationships below for more.
  • LANGUAGE LEARNING: Duolingo was first out of the gate with Duolingo Max, but you can, of course, have a conversation with ChatGPT in another language and ask it to correct your usage and pronunciation. also get a long way with the and now there are a slate of immersive language tools like Langua (more engaging than Duolingo), Glossarie.app, Speak,and TalkPal. Mondly VR and ImmerseMe add an immersive VR element. Speakable bills itself as your all-in-one language TA.

IMAGES, SLIDES, VIDEO, MUSIC and more Multimodal AI can both create and analyze sound, pictures and video.

  • IMAGES:Image Creator from Microsoft Designer still provides free access to the latest goodies in Dalle3.
  • ImageFX by Google and even better Imagen3 from DeepMind (also Google).
  • DALL-E by OpenAI (but not free for the best model)
  • Runway (based on Gemini): Note this new Act-1 feature that allows you to control animation with your own facial expressions.
  • Adobe Firefly especially for integrating images into photos
  • Stable Diffusion is good, cheap and open source
  • Midjourney is one of the best but can only be used through Discord.
  • Ideogram is another good tool that you can try for free.
  • Whisk allows you to use images as prompts for new images.
  • Generative AI by Getty Images creates stock images
  • SLIDES: Canva.com, Beautiful, Plus AI, Gamma and here is a review of seven more. Plus AI is an extension that allows you to make slides with AI in Google Slides. Napkin is also useful for slides as it creates infographics and visualizations from your text (without prompting).
  • Here is a video about how to make slides in Google Slides with Plus AI
  • Here is a video on How to make Slides with ChatGPT in Power Point.
  • VIDEO: Sora (from OpenAI with ChatGPT Plus), Runway (based on Gemini), Veo 2(from DeepMind), Flux, Movie Gen (from Meta) Kling (from Kuaishou Technology) and lots of APIs (that use ChatGPT but with layers of tools): here is a review of 14 of them. The best open source text to video seems to be HunyuanVideo from Tencent in China. You can find a good comparison video of the different tools side by side with the same prompt here. Here is great review and demo of ten or the tools that convert images to video. Genie 2 converts images into interactive virtual worlds/games.
  • MUSIC: Suno will convert your instructions into a new song. As always more specific instructions yield better results, and not that the real benefit is that you can customize for specific people and events. (Try creating a gratitude song for a friend in a style they like with the details of what they did.) Udio is another tool for this. 
  • AVATAR CREATOR: HeyGen (You can now also send an avatar to a Zoom meeting.)
  • AI COMPANIONS: Character.ai, nomi.ai, replika, elliq (focused on senior lonelliness) friend, EdSight talks to students as a way to help colleges hear student voices.

MINI MODELS and EDGE AI: These are smaller, faster and more specialized (often) OPEN SOURCE tools that you customize to live and run on your phone. Note that the ways to make an LLM better are model size (see Frontier models above), data set size and and the amount of training. Since it is not clear that larger more capable models will be cost effective, these faster smaller models (with more training) may end up being more useful. Apple Intelligence will test this idea. More smaller models are coming. 

  • Phi-3.5 from Microsoft comes in three sizes Mini, Small and Medium (3.8-41B parameters)
  • OpenELM is the Apple version that comes in four sizes (270M-3B parameters)
  • Gemma is the open source smaller model from Google also in several sizes

BROWSER EXTENSIONS: One way to become more familiar with how AI works is to add an AI extension to your browser. If you use Chrome, some good free extensions are Perplexity AI, in the Chrome store here) SciSpace (which does everything SCiSpace above does, but in your browser), Merlin AI, Bing, or Clipy AI: now every time you do a Google search, you will also get an AI response.

AGENTS: A chatbot can only chat with you, but an “agent” can plan and execute a series of tasks, like building you a website or finding information on your computer. Another use of agents (that is also about growing use of synthetic data) is this simulated hospital with AI agents as both patients and doctors, which allowed the AI doctors to gain experience (treating 10,000 patients) and “evolve” become better. LinkedIn has an agent that helps recruit job seekers (and also an AI jobs match tool). Claude how has “computer use” if you use its API and you can get instructions here. Devin is not quite there, but it gives a taste of what is to come. Early tools include Swarm from OpenAI and Asana. OpenAI has introduced Operator (Jan 23, 2025). Here is a demo (from Graham Clay) where Operator has been asked to write an essay in a GoogleDoc at human speed with edits.

Both OpenAI and Google now have tools called Deep Research. Google’s is not a reasoning model and leverages Google’s search capabilities. The OpenAI version, however is a combination or reasoning+agent. Here is a demo from Ethan Mollick.

You can find a complete list of AI products (tracked by Ithaka S+R) here

There is a great AI guide for students.