ChatGPT by OpenAI

ChatGPT runs on the GPT-5 model family. The everyday default is GPT-5.5 Instant, fast and capable for most tasks. GPT-5.5 Thinking is the frontier reasoning model that shows its plan before responding, so use it for complex strategy, analysis, or anything where you want to see the working. GPT-5.5 Pro (Pro tier) is the current flagship for the most demanding work.

Thinking Mode activates GPT-5.5 Thinking, which shows an upfront reasoning plan before it answers. It does not just give you a smarter answer, it shows you how it got there. That makes it easy to spot gaps, challenge the logic, or redirect it mid-thought. This is a different kind of output, not just a speed-versus-accuracy trade.

A one-time setup that tells ChatGPT your role, context, preferences, and how you want it to respond. It applies automatically to every conversation from that point on. You stop re-explaining your situation each session. Think of it as onboarding your AI to your world so it shows up informed every time.

Memory stores facts about you across conversations, your name, job, preferences, ongoing projects, and working style. It builds automatically as you chat, and you can view, edit, or delete anything it has saved. The more you use ChatGPT, the more useful it gets. Custom Instructions sets the baseline, Memory grows the relationship over time.

Projects are persistent workspaces that group related conversations, uploaded files, and custom instructions together. ChatGPT holds full context across every conversation inside a project, so there is no re-briefing between sessions. Each project holds up to 40 files and can carry its own instructions, separate from your global Custom Instructions.

File Library gives your uploaded documents, spreadsheets, and files a permanent home, accessible across any conversation without re-uploading. Before this, every upload was treated as one-off. Now your key reference files live in one place and can be pulled into any chat on demand.

ChatGPT Business and Enterprise plans support shared Projects, so a team can work from the same context, files, and AI setup. This shifts ChatGPT from a personal productivity tool to a team capability. Same brief, same knowledge base, same AI setup, shared across everyone who needs it.

Voice Mode lets you talk to ChatGPT and hear responses out loud, close to real dialogue. Advanced Voice Mode (Plus and above) is emotionally aware, responds with natural pacing and tone, and handles interruptions. It works on mobile, desktop, and Apple CarPlay. Free users get a daily preview, paid users get full access.

Record Mode captures live audio, meetings, brainstorms, voice notes, and produces an editable transcript in Canvas with an automatic summary. It is available on the desktop app (macOS) and rolling out to other platforms. The highest-ROI workflow is Record Mode followed straight away by Deep Research on any unknowns that came up.

Web Search gives ChatGPT live access to the internet, pulling current information into responses rather than relying on its training data alone. It kicks in automatically when a question needs up-to-date information, and you can trigger it yourself for anything that needs a current source.

Deep Research is a multi-step research agent. It plans an approach, searches dozens of sources, reads full articles and PDFs, synthesises the findings, and delivers a cited report. It takes 5 to 30 minutes and is closer to delegating to a research analyst for an afternoon than running a web search. You review and can change the plan before it runs, and interrupt at any time to refine the focus. It connects to SharePoint, OneDrive, Dropbox and Google Drive.

Company Knowledge connects ChatGPT to your internal apps, Slack, Google Drive, SharePoint, GitHub, HubSpot, and 60 or more others, and pulls organisation-specific answers with clear citations back to the source. It respects existing app permissions, so ChatGPT only sees what you are already authorised to access. No company data is used to train models.

Canvas is a side-by-side editing space where ChatGPT generates a document or piece of code and you refine it together in real time. Instead of copying outputs back and forth in chat, you work in a shared document with inline edits, version tracking, and section-by-section revision. Canvas can also build interactive apps and call APIs, not just edit text.

Data Analysis (formerly Advanced Data Analysis) lets you upload spreadsheets, CSVs, sales reports, or financial models and ask ChatGPT to analyse them in plain English. No formulas or coding required. It can clean messy data, run calculations, generate charts, spot trends, and summarise findings. One of the highest-ROI features for any manager dealing with regular reporting.

Images 2.0 is OpenAI's most capable image model (launched 21 April 2026). It applies a thinking step before generating, renders near-perfect multilingual text in images, produces up to eight coherent images from one prompt, and outputs at up to 4K resolution. It replaces DALL-E entirely and is much better for branded visuals, diagrams, and marketing assets.

Custom GPTs are tailored versions of ChatGPT, pre-configured with specific instructions, knowledge, capabilities, and tone. Build one once, use it repeatedly, share it across a team, or publish it in the GPT Store. They behave like a specialist assistant for a defined job. No coding required. Think of it as creating a custom AI role in your business.

Skills are reusable task bundles, instructions plus optional files or code, that teach ChatGPT exactly how to run a specific repeatable task. Instead of re-explaining your process each time, you build the Skill once and fire it with a slash command. Available in ChatGPT Business, Enterprise, and Edu plans, and exportable to other tools that support the Agent Skills format.

Apps (renamed from Connectors in December 2025) connect ChatGPT to external services and data so it can pull relevant context straight into responses. Some Apps give you interactive in-chat experiences, others are read/write connectors to tools like Google Drive, Slack, Notion, GitHub, HubSpot, Asana, Dropbox, and 60 or more others. Admins can also build custom connectors to any data source using MCP.

Agent Mode is a general-purpose agent that can reason, plan, browse the web, use tools, and take actions on your behalf, filling in forms, working with documents, building spreadsheets, and holding context across multiple tools. It is the shift from AI that advises to AI that acts. It asks for your approval before any sensitive action (submitting a form, sending an email, making a purchase), so you stay in control. It launched as "Operator" in 2025 and is now built into ChatGPT.

Workspace Agents are shared business agents for repeatable workflows across connected tools like Slack, Google Drive and SharePoint, for things like sales prep, ops workflows, customer support and reporting. They are available in Business and Enterprise workspaces. This is where ChatGPT starts operating as an organisational capability rather than a personal tool.

Create Task lets you schedule recurring or one-time operations, reminders, research queries, report generation, or any repeatable action, that ChatGPT runs at a set time, even when you are not in the app. It is the closest thing to a built-in AI scheduler. Up to 40 scheduled tasks per month on Plus.

Codex is OpenAI's autonomous coding agent, built into ChatGPT. You describe what you want built in plain English, Codex clones your GitHub repository into a sandbox, writes code across multiple files, runs your test suite, fixes failures, and opens a pull request. It works via the ChatGPT web interface, a desktop app (macOS and Windows), CLI, and IDE extensions for VS Code and Cursor. For non-technical users, the point is simple: it is a developer on call, working away while you focus on other things.