An AI assistant is a chat interface to a large language model. You ask in plain language, and it answers questions, drafts and rewrites text, summarizes documents, analyzes files and spreadsheets, searches the web and, in most cases, generates images. The individual plans of the main products cost about the same each month, so price rarely separates them.
What does separate them is how well they handle long documents, how they cite sources, how much use a plan allows before limits apply, and how closely they connect to the email, files and meetings you already use. For a business, three questions usually settle it: whether your data is used for training, which admin and retention controls come with the business plan, and which office suite the assistant is built into.
At a glance
| Tool | Best for | Starting price | Free trial |
|---|---|---|---|
 ChatGPT | Best for varied work that doesn't live inside one vendor's office apps | Free plan | No |
 Claude | Best for reading, writing and analysis that ends in a report, spreadsheet or deck | Free plan | No |
 Perplexity | Best for research where every claim needs a source you can open and check | Free plan | No |
 Gemini | Best for people whose mail, calendar and files already live in Google's apps | Free plan | No |
 Microsoft Copilot | Best for Microsoft 365 organizations that work in Excel, Outlook and Teams all day | Free plan | Yes |
5 of 5 tools
The ranking
In order, with what each tool is for and what it costs. Use the filters above to narrow the list.
Best for varied work that doesn't live inside one vendor's office apps
ChatGPT is OpenAI's AI assistant for people whose work spans writing, research, files, images and code. Its Work mode takes a task with a defined result and returns a document, spreadsheet, presentation or PDF to review.
Free plan
paid from $8/mo
Best for reading, writing and analysis that ends in a report, spreadsheet or deck
Claude is Anthropic's AI assistant for reading, writing and analysis that ends in a document, spreadsheet or deck. It returns Word, Excel and PowerPoint files, and paid plans include the Claude Code agent.
Free plan
paid from $17/mo
Best for research where every claim needs a source you can open and check
Perplexity is an AI answer engine for people who need to check where facts come from. It searches the web in real time and cites its sources in every answer, and its Research mode writes longer sourced reports.
Free plan
paid from $16.67/mo
Best for people whose mail, calendar and files already live in Google's apps
Gemini is Google's AI assistant for people whose email, calendar and files are already in Google's apps. It works with Gmail, Calendar and Drive inside a chat, and paid Google AI plans bundle it with Google One storage.
Free plan
paid from $4.99/mo
Best for Microsoft 365 organizations that work in Excel, Outlook and Teams all day
Microsoft Copilot is Microsoft's AI assistant, available as a standalone chat app and built into Word, Excel, PowerPoint, Outlook and Teams. It suits Microsoft 365 organizations and surfaces content a user already has permission to view.
Free plan
paid from $16.67/mo
Head to head
Feature-by-feature matchups between leading AI Assistants tools, with a clear verdict.
Methodology
We compare the tools on this page using what can be checked from the outside: published pricing and plan limits, documented features, and each vendor's own security, compliance and support documentation. This is desk research, not hands-on testing. Where something can only be judged by using a product, we say so rather than imply we measured it.
What we weigh here:
How the order is set:
Buyer's guide
Public benchmark scores move monthly and correlate loosely with whether an assistant is useful to you. The reliable test takes an afternoon: collect five tasks you regularly repeat, run them through two or three assistants on their free tiers, and compare. Teams that do this almost always pick a different product than the one the leaderboards suggested.
If your company runs Google Workspace, Gemini reads your Docs, Sheets and Gmail without an integration project. If you run Microsoft 365, Copilot does the same across Word, Excel, Outlook and Teams. That native access is worth more day-to-day than a modest quality edge.
Nearly every consumer tier here advertises generous limits and enforces quieter ones: message caps on the strongest models, slower queues at peak, smaller context windows on cheaper plans. This is not dishonest, but it does mean the $20 tier and the $200 tier can behave very differently on exactly the work you bought the tool for. Test at your real volume before you commit.
Check three things before rolling anything out: whether your inputs train the vendor's models by default, whether there is an enterprise agreement that turns that off, and what the retention window is. The answers differ meaningfully across the five products here, and they are far easier to establish now than after a team has pasted six months of customer data into a chat box.
FAQ
The standard consumer tier has settled at $20 per month across ChatGPT Plus, Claude Pro and Perplexity Pro, with Google AI Pro and Microsoft 365 Premium (which replaced Copilot Pro) slightly below at $19.99. Cheaper entry tiers have appeared too: Google AI Plus at $4.99 and ChatGPT Go at $8. Business tiers run $20–32 per user per month billed yearly, except Microsoft 365 Copilot at $21–30 on top of your existing Microsoft licensing. Heavy-use tiers (ChatGPT Pro, Perplexity Max, Claude Max) sit at $100–500 per month and are only worth it if you hit limits daily. Every product here has a free tier worth testing first.
Claude, in our editorial assessment. Its default register carries less of the recognisable AI cadence, which means shorter edit passes before the output is publishable. ChatGPT is close and more versatile if you also need images or data analysis. For research-heavy writing where claims need sources, pair either with Perplexity.
On free and consumer tiers, often yes by default, and this is the single most important thing to check before a team starts pasting customer data into one. Business and enterprise tiers of all five products here exclude your data from training, which is the main reason to buy them beyond admin controls. Verify the retention window too: excluding data from training is not the same as deleting it promptly.
Mostly. ChatGPT comes closest to covering every task competently, and for the majority of people one subscription is the right answer. The common exception is research: if you regularly need claims backed by checkable sources, Perplexity at $20 is a sensible second subscription. Beyond that, a second general assistant rarely earns its cost.
Accurate enough to be useful and wrong often enough that verification is not optional. All five state incorrect things confidently, and the failure mode is subtle: plausible, well-formatted and wrong. Assistants with live web search and citations (Perplexity, and the search modes of the others) are easier to check because you can follow the source. Treat any factual output as a draft to verify, not an answer to trust.
Write a review and help the next team choose. Every review is read by a moderator before it's published, and we never pay for reviews.
Write a review