An assistant and agent workspace for explanations, research, files, and connected work.
Conversation & research · Changing · Documentation checked 2026-10-04
ChatGPT Chat and Work; local Codex and the OpenAI API are separate surfaces.
Tools, usage, and controls vary by plan, client, region, rollout, and workspace policy.
What the platform is for
ChatGPT can help a beginner ask questions and turn approved context into a useful draft. Its current Chat and Work paths cover different scales of interaction. Start with a result you can inspect: an explanation, a sourced briefing, or a small editable document. Choose deeper reasoning or additional tools only when the task needs them.
Know the product boundary
ChatGPT web, desktop, mobile, local coding clients, and the API are distinct surfaces. Projects on web do not automatically expose a computer folder. Work cloud and local access differ. Enterprise administration is not the same as a personal-account setting. Current Enterprise custom-GPT guidance describes migration to plugins; verify the edition before following an older tutorial.
How the components connect
The task supplies the goal. Sources supply evidence. Project instructions and skills supply context or method. A model reasons over that material; search, file tools, plugins, and Browser extend what it can inspect or do. A generated file is a deliverable, while a scheduled task repeats a configured request. Memory helps continuity but should not be the only home for essential rules.
A beginner starting point
First ask for an explanation using public or invented context. Next attach one permitted document and check a summary against it. Then learn source verification and project organization. Only after that add a trusted connector, browser task, or schedule. The guides below provide an ordered path and two complete recipes.
Limits and data choices
Confident answers, citations, and attractive files can still contain mistakes. Check important claims and calculations directly. Scope connected identities and review consequential actions. Plan, surface, workspace policy, and rollout affect access. This edition does not exhaustively cover dots, Space, every creative mode, or every enterprise control; the inventory identifies the documented components taught here.
The technical path
For local development, continue to the Codex platform map. For integration engineering, use the OpenAI developer documentation linked from the official Learn site to study APIs, tools, authentication, evaluation, and deployment. Consumer subscriptions and API pricing / limits must be verified independently.
Feature map
Chat, Work, and model choices
Choose an interaction style and compare model or reasoning settings against a clear task.
Search and deep research
Find current information and build a report whose sources can be checked.
Files, analysis, and deliverables
Turn supplied documents or data into a reviewable analysis and an editable file.
Projects, instructions, and memory
Keep related sources together and distinguish deliberate project instructions from recalled context.
Images and multimodal interaction
Use images as inputs or generated drafts and understand the different voice and video boundaries.
Skills, plugins, apps, and MCP
Separate reusable instructions from external tools and the authority granted to connected accounts.
Scheduled tasks and event triggers
Repeat a proven task with an explicit schedule, input scope, and review destination.
Permissions, administration, and product boundaries
Audit workspace access, runtime permissions, and external actions before expanding an agent task.
Browser and computer use
Inspect a website or perform a bounded browser task with deliberate website and action permissions.
Complete workflows
From evidence to a beginner lesson
Create a sourced workshop lesson and an editable handout from approved material.
A weekly evidence digest with review
Repeat a public-source investigation and deliver a private draft digest.
Coverage and availability
Models, modes & conversation
Covered: Model and reasoning choices vary by client; Chat and Work explained.
Search, research & sources
Covered: Search and deep research; citation verification required.
Files, analysis & code
Covered: Supplied files, analysis, and generated-file review; supported formats vary.
Images, video, audio & voice
Not established: Images and supported desktop voice covered; full video / audio product parity not established.
Projects, instructions & memory
Covered: Projects and memory; local Codex memory is separate.
Editing surfaces & deliverables
Covered: Reviewable files and previews covered; every legacy Canvas or Space surface is not covered.
Agents, skills, plugins & MCP
Covered: Skills, plugins, apps, MCP, and Browser boundaries; dots is outside this edition.
Orchestration & scheduling
Covered: Scheduled tasks and event-trigger distinction; environment dependencies explained.
APIs, SDKs, deployment & evaluation
Separate product: Codex has its own map; developer APIs, SDKs, and deployment require separate docs.
Sharing, privacy & governance
Covered: Identity, runtime authority, workspace policy, and external actions distinguished.
Plans, devices & accessibility
Not established: Plan and surface relationships documented; exact quotas and provider accessibility behavior not established.
Sources and verification
Documentation checked 2026-10-04. Examples are illustrative unless described otherwise.
Using ChatGPT: Chat and Work · OpenAI
Model selection · OpenAI
Usage and billing · OpenAI
Web search and deep research · OpenAI
Work with files · OpenAI
Projects and chats · OpenAI
Memories across product surfaces · OpenAI
Image generation · OpenAI
Build skills · OpenAI
Plugins and supported surfaces · OpenAI
Build plugins and custom GPT migration · OpenAI
Scheduled tasks · OpenAI
Permissions and approval policies · OpenAI
Workspace administration · OpenAI
Get started with Work · OpenAI
Browser and computer-use boundaries · OpenAI
OpenAI API developer documentation · OpenAI