Futurepedia

The Complete Guide To ChatGPT

Most people open ChatGPT, type into the box, and stop there. That's a mistake.

Get exclusive savings on HubSpot Starter and ChatGPT Business, plus the tools to turn AI-powered discovery into real growth.
Get the limited time offer

Introduction

ChatGPT has quietly grown from a single chat window into a full ecosystem of products — one for quick answers, one for finished deliverables, one for talking out loud, one that works on your computer, one for connecting your tools, one for team workflows, and one for images. Each is built for a different kind of work. Most people are using maybe one of them.

This guide gives you the complete picture: what each product does, when to reach for it, and how they fit together. Scan the ecosystem table, find where you are, then go as deep as you want. You already know how to chat — this is everything you're missing.

Current Model Lineup

ChatGPT runs on different underlying models — think of them as engines, each tuned for a different job. On most paid plans you don't pick a model name directly; you pick how hard it thinks with a reasoning control, and ChatGPT selects the engine.

The current flagship is GPT-6. GPT-6 Astra powers ChatGPT Work and Codex, and shows up in Chat as GPT-6 Pro on higher tiers. Everyday chat still runs on the GPT-5.6 family, while GPT-6 is available separately for certain higher-end use cases and plans.

What you're doing
What you'll use
Notes
Everyday chat (Free & Go) GPT-5.6 Luna The default for Free and Go. "Think" also runs on Luna.
Everyday chat (paid plans) GPT-5.6 Sol + reasoning slider Choose Instant → Medium → High → Extra High for more thinking on harder tasks.
Hardest problems (flagship) GPT-6 Pro (powered by GPT-6 Astra) The top tier in Chat — on Pro, Business, and Enterprise. GPT-5.6 Sol Pro is the highest-capability GPT-5.6 option below it.
Inside ChatGPT Work & Codex Sol / Terra / Luna (+ GPT-6 Astra) Terra = balanced everyday work; Luna = fastest and lowest-cost; Sol = complex synthesis. Plus and up also get GPT-6 Astra here.
Live voice GPT-Live-1 (paid) / GPT-Live-1 mini (Free & Go) Powers the Live voice experience.
Images ChatGPT Images 2.5 Native image generation across tiers.

Which model should I use? The ChatGPT Ecosystem at a Glance

Which should I use? For most work — writing, research, analysis, everyday chat — the standard model at Instant or Medium reasoning is your default. When a task is genuinely hard (deep research, financial modeling, multi-step strategy), turn the reasoning up to High or Extra High, or step up to a Pro model (GPT-6 Pro on eligible plans). Inside Work, Terra handles most jobs; save the heaviest model for complex, ambiguous synthesis.

The ChatGPT Ecosystem at a Glance

ChatGPT isn't one tool — it's seven surfaces. Here's how they fit together before we go deeper.

 

Product

What It Is

Best For

Available On
ChatGPT (Chat) The core chat interface Quick answers, writing, research, analysis, Q&A All plans (web, desktop, mobile)
ChatGPT Work Agent for longer, multi-step work and finished deliverables File-heavy jobs, connected-tool workflows, automations Desktop on all plans (macOS/Windows); web & mobile on paid plans
ChatGPT Voice Live conversational interface Hands-free thinking, learning, brainstorming Web/iOS/Android across plans; agentic desktop Voice inside Work/Codex
Codex Agentic tool that acts on your computer Building tools, cleaning data, automations, coding All plans (usage varies by plan)
Plugins, Apps & MCP How ChatGPT connects to your tools and data Pulling context from Gmail, Drive, Slack, CRM; custom integrations Varies by plan; full MCP is beta on Business/Enterprise/Edu (web)
Workspace Agents Codex-powered agents for repeatable team tasks Standardizing org workflows across ChatGPT and Slack Business, Enterprise, Edu
ChatGPT Images Native image generation Visuals, mockups, marketing assets All tiers; web/mobile plus Work/Codex/desktop

 

ChatGPT: Where Most People Start

Chat is the front door — web, desktop, and mobile. It's where you ask questions, draft content, analyze documents, and think out loud. Everything else connects back to it. A few features quietly do most of the heavy lifting:

Search

ChatGPT can search the web in real time and cite its sources, so you're not stuck with stale answers. Trigger it yourself ("search the web for…") or let it decide when a question needs fresh data. Use it for anything time-sensitive — pricing, recent news, current stats.

Projects

Projects are self-contained workspaces with their own files, chat history, and custom instructions. Upload your brand guidelines, past reports, or SOPs once, and every chat in that project has that context automatically.

One caveat worth knowing: context lives in the project's knowledge base, not in your memory of past chats. If you want ChatGPT to use a document across chats, upload it to the project — don't just mention it once. (File limits vary by plan — Free is lower, paid plans allow more.)

Deep Research

Deep Research is a mini-agent inside ChatGPT that goes off for several minutes, pulls from dozens of sources, adjusts its approach, and compiles a cited report — the kind of literature review that used to eat an afternoon. Always review and verify before you act on it, but as a first pass it's a serious time-saver.

Memory & Personalization

ChatGPT builds a memory of your role, preferences, and ongoing work, and surfaces it without you re-explaining. You can view, edit, or delete memories in Settings, and set tone and communication preferences (formatting, warmth, response style) that apply across chats. Turn on temporary chats when you want a conversation left out of memory entirely.

✦ Try This Prompt

Search the web for the latest developments in [TOPIC] over the last 6 months and give me a 3-bullet summary of what actually changed and why it matters for [MY ROLE/INDUSTRY].

ChatGPT Work: Delegate the Whole Job

Chat is a conversation. Work is a delegation. Instead of going back and forth, you describe an outcome and Work plans the steps, does them, and hands back a finished deliverable — a real file, not text to reformat. It runs in the cloud on web and mobile, and on the macOS/Windows desktop app it can also work with your local files, desktop apps, and browser.

How to think about it: three levels

  • 1. Do the work (files → deliverable)

    Give Work your files and a clear outcome — "turn these three docs into a one-page brief" — and it produces the finished file.

  • 2. Connect the work (your real tools)

    Install Plugins to connect Gmail, Google Calendar, Slack, Drive, or your CRM, then reference them with @ in any prompt (@Gmail, @Slack). Now Work operates on your actual business context, not just what you remembered to upload. On Business/Enterprise plans this is where connected company data with source citations comes in.

For web and computer control

Work uses two documented browser surfaces: the Cloud browser (a remote browser it runs from Work on web/mobile) and the Built-in browser (inside the desktop app). When a task needs your existing Chrome profile, logged-in session, or open tabs, OpenAI points you to the Codex Chrome extension.

Best for vs. not for

Best For

Multi-step knowledge work, file-heavy tasks (Docs, Sheets, decks), recurring workflows you want automated, anything where the output is a real deliverable. 

Not For

Quick questions or short edits (use Chat), or work where you need to be in the loop on every decision.

Honest Limitation

You still own review and approval — Work drafts, you sign off. And local Scheduled Tasks need your computer on and the desktop app running; cloud tasks run independently, so use those for anything overnight.

See the difference:

  • In Chat: "Write me a 5-email launch sequence." → You get text to copy, paste, and format.
  • In Work: "Using the product brief in @Google Drive, write a 5-email launch sequence and save each as a separate Doc in our Campaigns folder." → You come back to five finished files.

✦ Try This Prompt

Review my @Gmail, @Google Calendar, and @Slack from the last 24 hours. Identify everything that needs my attention today — conflicts, deadlines, decisions I owe, overdue follow-ups. Give me a prioritized plan with source links, and draft replies where the next step is clear. Flag anything you couldn't access. Do not send anything.

Futurepedia_Header_ChatGPT Work

Guide to ChatGPT Work

From your first prompt to fully automated workflows — everything you need to get real work done with ChatGPT Work.

ChatGPT Voice: Think Out Loud

Voice is the live, conversational way to use ChatGPT — talk, interrupt naturally, and get spoken answers back. It's great for brainstorming on a walk, talking through a problem, learning a topic, or working hands-free. The umbrella product is ChatGPT Voice, and inside it you may see three experiences:

Guide to ChatGPT_voice
  • Live — the newest real-time experience (the most natural, full-duplex conversation).
  • Advanced — the previous real-time mode. On mobile it still carries video and screen-sharing that Live doesn't have yet.
  • Standard — the baseline voice experience.
Voice is available on web, iOS, and Android across plans, and there's an agentic desktop Voice inside Work and Codex on macOS/Windows — so you can speak, interrupt, and coordinate tasks using that experience's tools and permissions.

Honest limitations (worth knowing)

  • Live doesn't yet support camera/vision, video, screen sharing, Apps, or Plugins — those remain in Advanced on mobile. Pick the experience that has what you need.
  • Only one Voice conversation can run at a time.
  • You can shape tone, pace, response style, and preset voices — but there's no documented "emotion control" dial, so don't expect one.

Voice vs. Dictation

Use Voice for a live back-and-forth conversation. Use Dictation when you want to speak a prompt and get it back as editable text you'll review before sending. Different jobs.

✦ Try This Prompt (say it out loud)

I'm about to walk into a meeting about [TOPIC]. Quiz me on the three toughest questions I might get, then give me a sharp one-sentence answer for each.

Codex

Codex looks and feels like ChatGPT — a prompt box, plain English — but instead of just responding, it acts on your computer. Point it at a folder and tell it what you want: it organizes files, cleans data, builds spreadsheets and tools, controls a browser, and sets up automations. It's available on every ChatGPT plan (you download the desktop app), and there's a "for everyday work" mode that hides the technical detail.

 

 

What it's good for

  • Guide to ChatGPT_codex1

    For non-developers

    The sweet spot is building small, useful things that would otherwise mean hiring someone or wrestling with spreadsheets — cleaning a messy folder of sales logs into one report, building a simple internal dashboard, running a data analysis, automating a recurring task.

✦ Try This Prompt

Here's a folder of [messy files — e.g. sales logs, invoices, PDFs]. Organize them into clear subfolders, then build one spreadsheet that pulls everything together so I can see [WHAT YOU WANT TO SEE]. Before you start, list your plan and flag anything that looks inconsistent or missing.

  • Futurepedia_Header_Codex Prompts That Replace Busywork

    Codex Prompts That Replace Busywork

    5 copy-paste prompts for daily briefs, weekly summaries, data cleanup, and reusable Skills, no dev background needed.

    Access

Plugins, Apps & MCP: Connect ChatGPT to Your Tools

This is how ChatGPT stops working in a vacuum and starts working on your context.

Three layers, and the distinction matters:

  • Plugins are the discovery and packaging layer — the primary way to find workflow capabilities. A Plugin can bundle instructions/Skills, one or more Apps, and templates.
  • Apps are the actual integrations with services and data — Google Drive, Gmail, Slack, your CRM. (These were renamed from "Connectors" in December 2025 — same functionality, new name.)
  • MCP (Model Context Protocol) is the protocol behind custom Apps. Full MCP — including write actions — is currently a beta capability for Business, Enterprise, and Edu on ChatGPT web; mobile MCP Apps aren't supported yet.

How To Use It

Open the Plugins tab, connect the tool, then reference it with @ in any prompt. Once Gmail, Calendar, Slack, and Drive are connected, prompts get dramatically more powerful because the context gets richer.

A Note On Custom GPTs

If you've built or used Custom GPTs, know that they're now in retirement. Existing GPTs still run for now, but new creation is being turned off, and OpenAI's documented migration path is Plugins — not Workspace Agents. If you're setting something up fresh today, build it as a Plugin/Skill, not a new Custom GPT. (Team note: re-verify the current retirement dates before publishing.)

✦ Try This Prompt

Connect @Gmail, @Google Calendar, and @Slack, then give me a Monday-morning briefing: what's due this week, which threads are waiting on me, and which meetings I should prep for. Link every item back to its source.

Workspace Agents: Standardize It for the Whole Team

Where a Skill packages a workflow for you, Workspace Agents package a repeatable task for your whole organization. They're Codex-powered agents for the jobs a team runs over and over, and builders can attach tools, Apps, custom MCPs, Skills, files, schedules, Slack channels, and API triggers.

They run in ChatGPT and Slack (a Codex builder Plugin is in beta) and are available to Business, Enterprise, and Edu plans. The payoff is consistency: instead of ten people prompting ten different ways, everyone runs the same vetted workflow and gets the same standard of output.

Honest note: OpenAI's own documentation is inconsistent about whether Workspace Agents are labeled "research preview" or "generally available." Treat the product as live, but re-verify the label before you promote it.

✦ Try This Prompt

I want a Workspace Agent that runs our [RECURRING TEAM TASK — e.g. weekly pipeline review]. It should pull from [TOOLS], follow [OUR FORMAT/STANDARD], run every [CADENCE], and post the result to [SLACK CHANNEL]. Outline the setup before building.

ChatGPT Images: Visuals Without a Designer

ChatGPT Images 2.5 is native image generation built right into ChatGPT — describe what you want and get a usable visual, mockup, diagram, or marketing asset. It's available across tiers on web and mobile, and inside Work, Codex, and the desktop app.

It's ideal for first drafts and concepting: social visuals, simple mockups, illustrative graphics, "show me what this could look like." As with everything else, treat the output as a starting point you review and refine — then hand off to a designer for production polish when it matters.

Create a clean, on-brand [ASSET — e.g. social graphic / one-pager hero image] for [CAMPAIGN]. Style: [DESCRIPTORS]. Include [ANY TEXT]. Keep it simple and uncluttered, and give me two variations.

Real-World Applications

Here's what it looks like when the pieces work together.

Use Case 1: The Marketer Launching a Campaign

You're a marketing manager with a launch in three weeks and a small team.

  • 9-Sep-16-2026-04-06-17-4296-AM

    1. Chat (Search + Deep Research)

    Start with competitive and audience research. Get a cited brief on what's working in the market right now.

Use Case 2: The Founder Preparing to Pitch Investors

Early-stage, strong domain knowledge, limited bandwidth for pitch prep.

  • 10-Sep-16-2026-04-08-18-6695-AM

    1. Deep Research

    Run market sizing, competitive landscape, and recent funding activity — with citations, so your slides aren't just assertions.

Use Case 3: The Creator Turning One Video Into a Week of Content

Solo creator or podcaster who publishes consistently and loses hours to repurposing.

  • 9-Sep-16-2026-04-06-17-4296-AM

    1. Chat + Search

    Paste your transcript and validate which angles are landing in your niche right now; identify the strongest hooks.

100 Ways to Try ChatGPT Today

Ready to Get Started?

Breeze Assistant is your AI go-to-market expert, built into HubSpot. Start using these prompts today.

Get started free