Before you read: model names and version numbers change more often than anything else in AI, sometimes several times a month. This guide explains the ideas and leaves the model list to the vendors. Check the published date, and talk to Softwired if you'd like training built around today's line-up.
The short answer
A model is the AI "brain" behind a tool. ChatGPT, Claude, Gemini, Copilot, Grok and Meta AI are the apps, and each app runs its own models, which get new versions regularly.
A thinking mode (also called reasoning or extended thinking) makes the model work through a problem step by step before it answers. It is slower and better for hard questions.
A context window is how much the model can hold in mind at once: your messages, its replies and any files you add.
Models: the app versus the brain
Think of the app as the car and the model as the engine. Vendors swap in new engines regularly. OpenAI's model release notes for October 2026 show new models arriving and older ones retiring within weeks of each other, and Microsoft 365 Copilot now offers models from more than one company.
What this means for you:
- Most people can leave the picker on the default, often called Auto, and let the tool choose a suitable model.
- Paid plans usually include more capable models and higher limits.
- Results can shift when the model changes, so a prompt that worked last month may behave a little differently today.
- Some features work only with some models. OpenAI, for example, notes that some newer models are offered in ChatGPT Work and Codex but not in Chat.
Thinking modes
Switch on a thinking mode when accuracy matters more than speed. The table shows where each mode earns its place.
| Use a fast, everyday mode for | Use a thinking mode for |
|---|---|
| Quick questions and rewording | Multi-step maths, logic or planning |
| Short emails and summaries | Analysing a complex document or data |
| Brainstorming | Comparing options with many trade-offs |
| A customer reply you need in the next minute | Checking a calculation or a contract clause |
You'll see thinking modes under different names: a "Thinking" option in ChatGPT's model picker, extended thinking in Claude, and thinking or "deep reasoning" options in Gemini and Copilot. Some tools decide for themselves when to think harder. Thinking draws more on your plan's allowance, so save it for the tasks in the right-hand column.
Context windows
The context window is the AI's working memory for a single conversation. Everything counts towards it: your messages, its replies, pasted text and uploaded files. Every window has a limit, and the limit varies by model and plan. Sizes change often, so this guide leaves them out.
When a conversation gets long:
- The AI may lose track of details from early in the chat.
- Some tools summarise or "compact" older parts of the chat to make room.
- Answers can lose accuracy before the window is full.
The context window is separate from memory, which holds saved facts the AI keeps between conversations, and from projects, which hold files and instructions for every chat in the project. Our guides to AI memory and custom instructions and projects cover both.
How you'd actually use it
Pick the right mode
Drafting a quick reply to a customer? Use the default. Working out a roster for 14 staff across 3 sites with leave requests? Switch on thinking.
Start fresh when a chat drifts
If a long chat starts mixing up details, ask: "Summarise the key decisions and facts from this conversation in a short brief." Paste the brief into a new chat and carry on.
Put long-running background in a project
Add your style guide to a project once. Every chat in the project can draw on it, and you stop pasting it into each conversation.
Re-test important prompts after a model change
If your team relies on a prompt or skill for weekly reports, run it on a known example after each model update and compare the output with the last good run.
Frequently asked questions
Which model is best?
The best model depends on the task, and the answer changes often. For most work, start with the default or Auto setting and try a thinking mode for hard problems.
What is a token?
A token is the unit AI uses to measure text, roughly a piece of a word. Vendors measure context windows and usage limits in tokens.
Why did the AI forget what I said earlier?
Either the conversation grew long enough to squeeze out early details, or you're in a new chat that can't see the old one. Memory, projects or a short summary carry information forward.
What is a knowledge cutoff?
A knowledge cutoff is the date a model's training data runs to. Newer information comes from web search, if the tool has it switched on.
Related guides
- AI memory and custom instructions explained
- What is a project in ChatGPT, Claude and other AI tools?
- AI tool features explained: a plain-English guide for 2026
Put it to work with AI On Ramp
Picking the right model and mode for each job is a habit, and habits are easiest to build together.
AI On Ramp gets your team using AI with confidence, on their own work and across teams, while you keep its cost and risk under control. We set up the tools, then our consultants sit beside your most advanced AI users and build real in-house workflows with them. Your nominated AI Lead gets the governance tools and our ongoing support to keep the business current as the tools change. We bring deep technical expertise and decades of business change management, and we work face to face from Brisbane and along Australia's east coast. See how AI On Ramp works.
Want to talk through how this applies to your team? Book a call with Softwired.
