Lesson 1: What Gemini Is: Models, Limits and This Course

Welcome: What This Course Is (and Is Not)

This course teaches you to use Google Gemini at work. You will not write any code. Each level ends with a task you do in Gemini and a way to check the result.

This course is independent of Google. Google does not write, review or endorse it, and it does not copy Google's documentation. Gemini changes almost every month: models get new names, limits change and features are renamed. For example, on 30 September 2026 Google announced that a new feature called Skills is replacing Gems. So the course checks each feature against Google's current pages, and it tells you when something is in preview or could not be confirmed. Learn the same habit: what your own screen shows today counts for more than anything you read, including this course.

What Gemini Is

Gemini is Google's family of generative AI models, plus the products built on them. It reads what you type or upload and writes new content in reply: text, summaries, tables, images and more.

Think of it as a fast, well-read assistant on their first day. It is capable and quick, but it has no context about your company and it can be confidently wrong. You give it direction, and you check its work.

You will meet Gemini in two main places:

1. The Gemini app. A chat assistant at gemini.google.com, on mobile and in desktop apps. Use it for open-ended questions, drafts, uploads and research.
2. Gemini inside Workspace apps. A side panel and in-app tools in Gmail, Docs, Sheets, Slides, Drive and Meet. Here Gemini works on the email, document or file you have open, so you don't have to switch tabs.

Which of these you can use depends on your plan and on your administrator's settings. Level 2 covers this in detail.

Models: The Engines Behind the Chat

A model is the AI engine that produces the answer. Right now Google's help pages describe a range of Gemini models:

• Flash-Lite: the fastest and most cost-efficient model, for quick answers.
• Flash: balances speed and quality.
• Pro: for more complex reasoning.
• Argon: Google's newest top-tier model. 'Gemini 4 Argon' began rolling out on 13 October 2026 to Google AI Pro and Ultra subscribers.

Two controls affect which model answers you:

• Auto model selection. Gemini sends each prompt to the model it judges best. Most prompts go to Flash-Lite, and harder ones may go to Flash or Pro. Since a rollout that began on 9 October 2026, free users with no plan see only 'Auto' in the model menu.
• Thinking level (Low, Medium, High). On paid plans you can choose how deeply the model reasons. A higher level is more thorough, but it is slower and uses more of your allowance.

Usage Limits: Why Gemini Sometimes Stops Helping

Gemini has usage limits. For personal Google Accounts, the limits are now compute-based: what counts is how much work your prompt takes, not simply how many prompts you send. The limit refreshes every 5 hours until you reach a weekly limit. These use up your allowance faster:

• More advanced models and higher thinking levels
• Long chats
• Deep Research and media generation (images, music and, most of all, video)

Work and school accounts follow different rules, which depend on the Workspace licence. Your administrator can ask for higher limits. In practice, use the lighter options for quick tasks and save the heavy features for work that needs them.

Task 1: Find Your Own Models and Limits

Do this now. Your account is what counts.

1. Open the Gemini app and sign in with the account you use for work.
2. Click the model selector at the bottom of the chat box.
3. Write down every option shown. If you see only 'Auto', write that down too.
4. Write down any label such as preview, beta, experimental or new. These are not final. They can change or disappear, so don't build a team process around them yet.
5. Open Settings and look for Usage limits. Record what it shows and whether auto model selection is on.
6. If you use a work account and some options are missing, record that. Your administrator controls them, and Level 2 explains how.

Check: you have a short note listing your models, any preview labels, and your limit settings, with today's date on it.

Task 2: Run a First Chat and Check It

Gemini can state wrong facts fluently. Make checking a habit from the first chat.

1. Choose a question from your own job that has a verifiable answer. Example (HR): "What is the current statutory minimum annual leave entitlement for full-time workers in the UK?"
2. Send it and read the answer.
3. Pick one specific claim in the answer, such as a number, a date or a name.
4. Find one trusted source for it: an official government page, your company's policy, or a vendor's own documentation. A blog or forum post does not count.
5. Compare the two. Mark the claim as matches, differs or can't find.
6. If it differs, the trusted source wins. Note the mismatch.

Check: you have one claim, one source link and one verdict. Level 4 builds this into a full method.

Key Takeaways

• This course is independent of Google and checks features against Google's current pages.
• Gemini appears in the Gemini app and inside Gmail, Docs, Sheets, Slides, Drive and Meet.
• Models range from Flash-Lite (fast) to Pro and Argon (deeper reasoning). Free users now get 'Auto' only.
• Personal limits are compute-based: they refresh every 5 hours up to a weekly cap. Heavy features use up the allowance faster.
• Treat anything labelled preview as subject to change.
• Check at least one claim in every answer against a trusted source.