The 2026 AI landscape

Free
0/10 steps visited

Step 4 of 10

Fast vs thinking

Read · ~5 min

Fast vs thinking

Open ChatGPT, Claude or Gemini today and you'll usually see more than one mode — not just "the model."

The names keep changing (Instant, Thinking, Pro, adaptive thinking, 3.1 Pro). The idea underneath stays put:

Mode (plain English)What it's forTrade-off
Fast / everydayDrafts, rewrites, back-and-forth, quick questionsCheaper, quicker; less depth on hard problems
Thinking / deepMulti-step reasoning, tricky analysis, harder codingSlower, eats more of your plan limit
Deep Research (a separate product in the app)Reports pulled from many sources, with citationsFew runs a month on Plus-tier plans; can take minutes

Same question, different mode

Fast chat: "Summarise this email in three bullet points." → Grand. Do it right here.

Thinking chat: "Here are three supplier quotes as messy PDF text — compare total cost of ownership over two years and flag the risks." → Worth the slower mode, if your plan includes it.

Deep Research: "Produce a sourced overview of the UK rules for X, for a client memo." → Not the same as pasting that into normal chat. It's a proper job the app goes off and runs — with monthly caps on most £20 plans.

I'll admit it: I burned Deep Research runs on stuff I could've cracked in fast chat, purely because the heavy mode felt more professional. Don't. Save the big modes for work that genuinely needs them — your caps are real and they run out.

Rules of thumb (no benchmark wars)

  • Default to fast for any language job you can judge in ten seconds.
  • Switch to thinking when you've already tried twice in fast mode and it's still missing the point.
  • Use Deep Research when you actually need sources and structure, not a one-paragraph guess.
  • Watch the limit counter in your settings — Plus plans often cap research runs per month.

We're not ranking which company "wins." We're stopping you swinging a sledgehammer mode at a thumb-tack job — and then wondering why your subscription feels so dear.

Continue — the subscriptions you're already paying for, including ones buried inside Microsoft 365.

Warning

Real power. Educational use only.

What we teach you to build is genuinely powerful — uncensored assistants, agents, and automations on your own hardware. In the wrong hands, that is as dangerous as malicious code in the wrong hands. We do not teach illegal, malicious, or harmful use. You are responsible for what you deploy.

See what we mean →