Your guide to Claude, ChatGPT, and Copilot — what each one does best, and when to use which.
This is one of the most common sources of confusion, and it's totally understandable. Here's the short answer: Copilot is built into Microsoft Office by Microsoft. The Claude and ChatGPT add-ins are separate tools installed on top of Office by Anthropic and OpenAI respectively. They appear in a sidebar inside Excel or PowerPoint, but they are running on different servers and require their own subscriptions.
One more wrinkle: there are unofficial "ChatGPT for Excel" add-ins in the Microsoft Marketplace (made by third parties, not OpenAI) that look similar to the official one. The official OpenAI add-in is simply called "ChatGPT" in the Microsoft Marketplace and requires a ChatGPT Plus, Team, or Enterprise account. If you had to enter a separate API key to set yours up, you may have the unofficial version.
All three tools are appropriate for business use — but they have different data privacy implications. See the cards below for details.
| Situation | Copilot | Claude Add-in | ChatGPT Add-in |
|---|---|---|---|
| File saved to OneDrive | ✓ Full Agent Mode | ✓ Works fine | ✓ Works fine |
| File saved locally (not OneDrive) | ✗ Agent Mode limited | ✓ Full capability | ✓ Full capability |
| Data must stay in M365 tenant | ✓ Best choice | ✗ Goes to Anthropic | ✗ Goes to OpenAI |
| Complex formula explanation needed | OK | ✓ Best | Good |
| Build charts / visual output fast | ✓ Best (native) | Good | Good |
| Cross-reference with Teams / Outlook | ✓ Only option | ✗ Not possible | ✗ Not possible |
| Write VBA macros from within Excel | Limited | ✓ Best | ✗ Beta limitation |
No tasks match your search. Try a different keyword or browse a category.
Copy-and-paste prompt templates to get great results from each AI tool. Replace the [highlighted placeholders] with your own details.
Not all AI models are equal — even within the same tool. This page explains which model to choose inside Claude and ChatGPT for your task, when to turn on Web Search, and when Extended Thinking is worth the extra wait. Copilot doesn't have a model picker, so no decisions needed there.
The fastest, lightest model. Best for simple tasks where speed matters more than depth. Uses the least of your monthly limit.
Your everyday workhorse — handles 90%+ of tasks with strong reasoning and fast responses. If you're unsure which to pick, start here.
The deep thinker. Reserve for tasks that genuinely need it — it uses more of your monthly limit. Also powers the Claude for Excel and PowerPoint add-ins automatically.
⚠️ Watch your limit: Opus uses significantly more of your monthly token allowance than Sonnet or Haiku. Use it intentionally — don't leave it selected for routine tasks.
ChatGPT simplified its model lineup significantly in early 2026. There are now just two modes — Instant and Thinking — plus an auto-routing option that chooses for you.
The fast, capable default for everyday work. ChatGPT will automatically switch to Thinking for complex requests when you leave it on Instant — so it's a smart auto-pilot.
Slower but more thorough. Reasons step-by-step before answering. Use for hard problems where you need the most accurate answer, not the fastest one.
Lets the AI search the internet in real-time before responding. Without it, both Claude and ChatGPT answer from their training data, which has a knowledge cutoff.
Makes the AI pause and reason through a problem step-by-step before giving you an answer. Responses are slower but significantly more accurate for hard problems.
Every AI conversation has a context window — think of it as the total amount of information the AI can hold in its memory at once. As a conversation gets longer, it fills up. When it gets too full, the AI may start forgetting earlier parts of the conversation, give shorter or lower-quality answers, or behave inconsistently. Claude has introduced "Infinite Chats" which largely reduces this problem — but ChatGPT still has limits, and very long Claude sessions can still degrade. The fix is simple: start a fresh conversation and paste in a brief summary of what you were doing.