AI Assistants: ChatGPT, Claude, Gemini & Copilot

Four AI assistants dominate everyday use: ChatGPT, Claude, Gemini and Microsoft Copilot. They look similar: a chat box, a model picker, buttons to attach files. Under the surface they come from different companies, run on different models and fit into different tools.

In this lesson you will learn who makes each assistant, what they have in common, where they differ, and a simple, fair way to choose the right one for your own work.

The four main assistants

A few details are worth knowing:

• ChatGPT made AI assistants popular when it launched in November 2022. OpenAI releases new GPT models several times a year, so check openai.com for the current names rather than relying on a version number from any tutorial.
• Claude comes in sizes. In late 2026 the current family is Claude Fable 5.1 (the most capable, for demanding reasoning and long, multi-step work), Claude Opus 5.5 (the recommended starting point for most work), Claude Sonnet 5.5 (a balance of speed and intelligence) and Claude Haiku 5.5 (the fastest, for high-volume simple tasks). All four accept up to 1 million tokens of context.
• Gemini also comes in sizes. Google's developer documentation lists Gemini 4 Argon as its flagship reasoning model and Gemini 3.8 Flash as its most capable Flash model, with lighter "Flash-Lite" models for cheap, fast tasks.
• Microsoft Copilot is a family of products. The free Copilot app and Copilot Chat give web-grounded chat. With a Microsoft 365 Copilot licence, an organisation gets Copilot inside Word, Excel, PowerPoint, Outlook and Teams, grounded in the user's own emails, files and meetings. GitHub Copilot is a separate coding assistant.

Besides these, there are open-weight models, whose trained weights anyone can download and run on their own computer or server. Meta's Llama family is the best-known example, and the Indian company Sarvam AI has released open models trained for Indian languages. Open-weight models need technical skill and good hardware, but your data never leaves your own machines.

What they have in common

All four assistants can:

• chat in many languages and remember the conversation within a chat;
• read files you upload, such as PDFs, spreadsheets and images;
• search the web and show links to sources (when the feature is on);
• write and explain code;
• save standing instructions or preferences so you do not repeat yourself.

Most of them can also talk with you by voice and create or edit images, though these features vary the most between assistants and plans.

Exact features depend on the plan (free or paid, personal or work account) and change often. Because the skills in this course (clear prompts, giving context, checking answers) work in every assistant, you can switch tools without starting again.

How they differ

The differences that matter most in practice are:

• Where they live. If your office runs on Microsoft 365, Copilot is already inside your email and documents. If your college uses Google Workspace, Gemini fits naturally. ChatGPT and Claude are strong stand-alone assistants that also connect to other tools.
• Data handling. Work and education accounts usually come with stronger privacy promises than free personal accounts. Some consumer plans may use your chats to improve models unless you switch this off. You will study this in Privacy, Company Data & the DPDP Act.
• Strengths on your tasks. One assistant may write Hindi more naturally, another may handle your spreadsheet better. Published rankings change every month and rarely test your kind of work.
• Cost and limits. Free plans limit how many messages or which models you can use.

Choosing fairly: test with your own tasks

The best way to choose is a small, fair test. Pick five tasks you really do, give exactly the same prompt to each assistant, and score each answer from 1 to 5 on accuracy, usefulness and how much editing it needed. Record your results in a simple format such as this:

{
  "task": "Summarise a 3-page circular about exam rules into 5 bullet points for students",
  "prompt_used": "Summarise the attached circular into 5 short bullet points for first-year students...",
  "scores": {"accuracy": 4, "usefulness": 5, "editing_needed": 3},
  "notes": "Missed the rule about calculators, otherwise clear."
}

Then a few lines of Python turn the scores into a fair comparison. The assistant names here are just labels for your own test results:

scores = {
    "Assistant A": [4, 5, 3, 4, 4],
    "Assistant B": [5, 4, 4, 3, 5],
    "Assistant C": [3, 4, 4, 4, 3],
}
tasks = ["summary", "email", "Excel formula", "Hindi letter", "code fix"]

for name, marks in sorted(scores.items(), key=lambda item: -sum(item[1])):
    best = [tasks[i] for i, mark in enumerate(marks) if mark == max(marks)]
    print(f"{name}: average {sum(marks) / len(marks):.1f}, best at {', '.join(best)}")

Assistant B: average 4.2, best at summary, code fix
Assistant A: average 4.0, best at email
Assistant C: average 3.6, best at email, Excel formula, Hindi letter

Notice that the overall winner is not the best at everything. Many people end up using one assistant for most work and a second one for particular jobs.

Common mistakes

• Choosing by hype or a single viral post. Test your own tasks.
• Using a personal account for office data when your organisation provides a work account with proper data protection.
• Memorising version numbers. Models change every few months; learn the skills, then check what is current.
• Assuming all features are free. Check what your plan includes before building a habit around a feature.

Key points

• ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google) and Microsoft Copilot are the main assistants.
• The current Claude family is Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5; Gemini's flagship is Gemini 4 Argon.
• Copilot is built into Microsoft 365 apps for licensed users; GitHub Copilot is a separate coding tool.
• All assistants share core features, but plans, data policies and integrations differ.
• Choose by testing the same prompts on your own real tasks and scoring the answers.

scores = {
    "Assistant A": [4, 5, 3, 4, 4],
    "Assistant B": [5, 4, 4, 3, 5],
    "Assistant C": [3, 4, 4, 4, 3],
    "Assistant D": [5, 3, 5, 4, 2],
}
tasks = ["summary", "email", "Excel formula", "Hindi letter", "code fix"]

for i, task in enumerate(tasks):
    top = max(marks[i] for marks in scores.values())
    winners = [name for name, marks in scores.items() if marks[i] == top]
    print(f"{task}: {', '.join(winners)} ({top})")

summary: Assistant B, Assistant D (5)
email: Assistant A (5)
Excel formula: Assistant D (5)
Hindi letter: Assistant A, Assistant C, Assistant D (4)
code fix: Assistant B (5)