You're Probably Using the Wrong AI Model. Here's Where to Change It.

Most people never touch the model dropdown. The app happily leaves you on a default that is often the fast or cheap tier, not the one that handles nuance best. Before you rewrite a prompt for the fifth time, open the selector and try the strongest option your plan allows. Same words in, noticeably different answers out.

Labels move whenever these companies ship a redesign, but the habit is the same: look for the model or mode name near the top of the chat (or in a menu tied to the input), click it, and switch. The table at the bottom is a cheat sheet for what to aim for as of early 2026.

Microsoft Copilot

Use the web app at copilot.microsoft.com or the sidebar inside Word, Outlook, and other Microsoft 365 apps. Your organization may show different names, but you are looking for the control that sets response style or the specific model.

1. Open Copilot in the browser or from the app you are working in.
2. Find the dropdown near the message box (sometimes labeled Balanced, Creative, Precise, or similar).
3. For drafting and creative work, Creative usually beats Balanced. If you see tiered GPT builds or a Think deeper path, use that when the answer actually matters.

This page does not use redrawn or "fake browser" pictures. For a real screen capture of Microsoft Copilot’s model menu (Auto, Quick response, Think deeper, More), open Microsoft Copilot: stronger models than Auto.

ChatGPT

On chatgpt.com, the active model is shown in the header of the conversation, not buried in settings. Free accounts see fewer choices; paid tiers unlock the full list.

1. Open or start a chat.
2. Click the model name at the top center of the thread (often "ChatGPT" with a caret).
3. For planning, math-heavy work, or anything that kept failing in the fast mode, pick Thinking (or the equivalent reasoning option). Use the standard mode when you need speed.

For more on when reasoning modes help, see reasoning models vs standard chat.

Claude

On claude.ai, the family (Sonnet, Opus, etc.) is visible from the chat shell. Extended thinking is toggled from the same area rather than a separate settings deep link.

1. Open Claude and start or select a conversation.
2. Use the model picker at the top of the chat or in the sidebar to see the current model.
3. Sonnet is the usual default for everyday work. Step up to Opus for long documents or hard reasoning. Turn on Extended thinking when you want the model to work step by step before it answers.

Gemini

On gemini.google.com, branding and model names sit high on the page. Advanced or Ultra tiers unlock heavier models if your Google account has them.

1. Open Gemini in the browser.
2. Check the model or product label at the top (Flash vs Advanced, etc.).
3. Flash-class models are fine for quick questions. For synthesis or research, use the strongest Gemini your subscription allows. Deep Research-style flows, when present, usually live off the + or tools menu next to the input.

At a glance

Tool	Stronger option (early 2026)	Where it usually lives
Copilot	Creative, Think deeper, or newest GPT build under More	Dropdown by the chat input
ChatGPT	Thinking / reasoning mode vs standard	Model name at top of chat
Claude	Opus + Extended thinking when needed	Top of chat or sidebar
Gemini	Advanced tier; Deep Research when available	Top of page; tools near input

Interfaces will change again. If you cannot find the control, search the help docs for "model" or "thinking" for that product, or skim how the major tools compare. The win is not memorizing pixels; it is remembering to check the model before you blame the prompt.

Explore AI 101 Prompt Library