ChatGPT

ChatGPT is a conversational artificial-intelligence assistant developed and operated by OpenAI. It was publicly released on November 30, 2022, as a free research preview based on a conversationally fine-tuned GPT-3.5 model. Its central innovation was not simply the underlying language model, but the productization of that model into an accessible dialogue interface in which users could ask follow-up questions, refine instructions, request transformations of text, and receive responses tailored to an ongoing conversation. The launch rapidly broadened public awareness of generative AI and helped trigger a large increase in investment, experimentation, and competition across the technology industry. The service began primarily as a text chatbot but expanded into a multimodal assistant. Depending on plan, region, device, and model availability, users can submit text, images, files, and voice prompts, while ChatGPT can return written answers, spoken responses, generated images, structured analyses, code, tables, and other outputs. Common uses include drafting and editing, translation, summarization, tutoring, brainstorming, software development, spreadsheet and data analysis, document review, trip and project planning, and research assistance. Features such as memory, projects, tasks, Canvas, custom GPTs, file uploads, data analysis, voice mode, and image generation have progressively turned the product from a general chatbot into a broader personal and professional workspace. OpenAI operates ChatGPT through a freemium model. A limited free tier provides access to selected models and tools, while paid individual plans have included Plus and Pro. Organizational offerings have been marketed through Team or Business and Enterprise plans, with additional administration, security, workspace, connector, and data-governance capabilities. Plan entitlements change as OpenAI introduces new models and reallocates usage limits. The product is available through the web, iOS and Android applications, desktop software, and selected integrations. ChatGPT Search adds web retrieval and source links for more current answers, while Deep Research performs longer multi-step research tasks and produces documented reports. Agent capabilities have extended the service toward computer-use workflows, coding assistance, and other tasks that can be executed across tools or virtual environments. ChatGPT also functions as a distribution layer for OpenAI’s model families. Major generations and capabilities have included GPT-4, GPT-4o, reasoning models in the o-series, native image generation, advanced voice interaction, and specialized agents. The GPTs system allows users to configure task-specific versions of ChatGPT using instructions, knowledge, and tools, and the GPT Store provides a discovery and distribution mechanism for publicly shared custom assistants. In 2025 and 2026, OpenAI increasingly positioned ChatGPT as an agentic platform capable of searching, synthesizing, interacting with websites, working with connected applications, assisting with code, and supporting shopping or other transactions. The brand’s importance is accompanied by substantial criticism. ChatGPT can produce fluent but false statements, reflect biases in its training data, and provide unreliable reasoning or citations. Its use has raised concerns about academic misconduct, misinformation, privacy, copyright, malicious code, emotional dependence, and the effects of automation on knowledge work. OpenAI has responded through model updates, moderation systems, privacy controls, safety evaluations, age-related protections, provenance metadata for generated images, and changes to problematic releases. Even with those controls, ChatGPT remains a probabilistic system whose outputs require human review, especially in medical, legal, financial, safety-critical, and other high-stakes contexts.

ChatGPT emerged from OpenAI’s work on large language models and instruction-following systems. Before the product launch, OpenAI developed GPT models that could generate and transform text, but the public typically encountered them through demonstrations, research releases, or developer products rather than through a single consumer-facing assistant. ChatGPT combined a GPT foundation model with conversational fine-tuning, supervised examples, reinforcement learning from human feedback, and safety mechanisms designed to make the system more useful and less likely to produce disallowed outputs. OpenAI released ChatGPT on November 30, 2022, as a research preview using a conversationally optimized GPT-3.5 model. The product was intentionally simple: users entered prompts into a chat interface and could continue a thread with follow-up questions. ChatGPT’s ability to preserve conversational context, revise answers, explain concepts, write in different styles, and assist with code made it accessible to people without machine-learning expertise. Its rapid adoption transformed generative AI from a specialist topic into a mainstream consumer technology category and accelerated competition among technology companies, search providers, productivity-software vendors, and model developers. In early 2023, OpenAI introduced ChatGPT Plus and made GPT-4 available to subscribers. This established the product’s freemium commercial structure: a broadly accessible free service combined with paid access to higher-capability models, larger limits, faster service, and new features. OpenAI then expanded ChatGPT beyond desktop web use. An iOS application launched in May 2023, followed by Android availability, while desktop software and voice input broadened the service’s role as an everyday assistant. The product’s scope expanded substantially during 2023 and 2024. Plugins allowed ChatGPT to connect with external services and specialized tools, although OpenAI later shifted much of that ecosystem toward GPTs, connectors, and native capabilities. In November 2023, OpenAI introduced GPTs, enabling users to create custom assistants for defined tasks. The GPT Store followed in January 2024, creating a marketplace-like layer in which custom assistants could be discovered and shared. Memory controls, file analysis, data-analysis tools, Canvas, projects, and tasks gradually repositioned ChatGPT as a persistent work environment rather than a single conversational model. The launch of GPT-4o in May 2024 marked a major multimodal phase. OpenAI emphasized faster interaction and improved handling of text, images, and audio, while expanding selected capabilities to free users. Voice conversations, image understanding, screen and file interaction, and later image generation made ChatGPT useful across more forms of input and output. OpenAI also introduced or expanded model selectors and reasoning-oriented models, allowing the product to serve both fast everyday interactions and more deliberate problem-solving tasks. ChatGPT became more explicitly connected to the live internet with the development of ChatGPT Search. First rolled out in late 2024 and subsequently broadened, Search allowed ChatGPT to retrieve current information and cite web sources. This moved the brand into direct competition with search engines while also creating new relationships and tensions with publishers, websites, and content owners. Deep Research, introduced in 2025, extended this direction by performing multi-step searches and synthesizing results into structured reports. During 2025, OpenAI also pursued agentic functionality. Operator demonstrated browser-based computer use, including interaction with forms and websites, and was later integrated into ChatGPT agent mode. Codex brought cloud-based coding-agent capabilities into the wider ChatGPT ecosystem. These developments shifted the product proposition from answering questions toward completing sequences of actions under user supervision. Shopping and commerce features similarly positioned ChatGPT as a conversational discovery and transaction layer. By 2026, ChatGPT had become a broad family of consumer and organizational services. OpenAI offered free, paid individual, business, and enterprise plans; continued to expand memory and agent features; introduced ChatGPT Health with additional privacy controls; and tested advertising in the free experience. The brand remains active and rapidly evolving. Its history is defined by short product cycles, frequent model changes, widening multimodality, and a continuing attempt to balance mass accessibility with reliability, privacy, safety, copyright, and commercial sustainability.

1. 2026ChatGPT Health U.S. rollout

   A dedicated health and wellness experience began launching to U.S. adults on web and iOS with additional privacy and isolation controls.

   Source
2. 2026International advertising pilot expansion

   OpenAI expanded ChatGPT ads testing to several additional countries while maintaining restrictions around minors and sensitive topics.

   Source
3. 2025Deep Research introduced

   Deep Research added an agentic research workflow capable of planning searches, evaluating sources, and producing documented reports.

   Source
4. 2025Operator integrated into ChatGPT agent

   OpenAI moved browser-use functionality from the standalone Operator preview into ChatGPT agent mode.

   Source
5. 2024GPT Store launches

   The GPT Store began distributing publicly shared custom assistants and featured creations from users and partners.

   Source
6. 2024GPT-4o and broader multimodality

   GPT-4o expanded text, image, audio, voice, and vision capabilities and brought selected tools to the free tier.

   Source
7. 2024Memory becomes a core product control

   OpenAI expanded memory features and controls so ChatGPT could use saved preferences and, for eligible users, relevant prior conversation context.

   Source
8. 2024ChatGPT Search introduced

   OpenAI added web search with links to relevant sources and progressively expanded availability across plans and regions.

   Source
9. 2023ChatGPT Plus launches

   OpenAI introduced a paid subscription with improved availability, faster responses, and access to newer capabilities.

   Source
10. 2023GPT-4 arrives in ChatGPT Plus

    GPT-4 became available in ChatGPT Plus, expanding the assistant’s reasoning and generation capabilities.

    Source
11. 2023Official mobile expansion begins

    OpenAI launched ChatGPT for iOS and subsequently expanded to Android, synchronizing chats across devices and adding voice input.

    Source
12. 2023Plugins and external tools

    OpenAI introduced plugins that enabled ChatGPT to use selected browsing, code, and third-party service workflows.

    Source
13. 2023GPTs introduced

    Users gained the ability to build customized versions of ChatGPT with tailored instructions, knowledge, and capabilities.

    Source
14. 2022Public launch as a research preview

    OpenAI released ChatGPT on November 30, 2022, as a free conversational research preview based on GPT-3.5-era technology.

    Source

A general-purpose, multimodal AI assistant and increasingly agentic productivity platform positioned for consumers, professionals, developers, educators, and organizations.

• ChatGPT FreeConsumer AI assistant2022

  The no-cost entry point to ChatGPT. The free tier is designed for everyday questions, writing, learning, planning, coding, file and image interactions, and selected access to newer models and tools. Limits, model availability, and feature access vary with demand and OpenAI’s product policies. The tier is central to ChatGPT’s mass-market distribution strategy and has supported the brand’s rapid adoption beyond enterprise and developer audiences.

• ChatGPT PlusPaid consumer subscription2023

  A paid individual plan introduced in February 2023. Plus has typically provided expanded message and tool limits, priority access, earlier availability of new capabilities, and access to higher-capability or reasoning-oriented models. Its exact benefits have changed as OpenAI has reconfigured model access and introduced features such as voice, image generation, Search, Deep Research, projects, tasks, and custom GPTs.

• ChatGPT ProPremium consumer subscription2024

  A higher-priced individual plan introduced in December 2024 for users seeking substantially greater access to advanced reasoning models, voice, Deep Research, image and video-related capabilities, and agentic tools. OpenAI describes the plan as providing the highest level of access among individual tiers, subject to abuse-prevention guardrails and changing product availability.

• ChatGPT Business and EnterpriseOrganizational AI workspace2023

  Organizational versions of ChatGPT that provide shared workspaces, administration, identity and access controls, enhanced privacy settings, connectors to business information, and access to tools such as data analysis, custom workspace GPTs, Deep Research, and coding agents. Enterprise is aimed at larger organizations requiring governance and support at scale, while Business is a self-serve workspace for teams.

• ChatGPT SearchAI search and retrieval2024

  A web-search capability integrated into ChatGPT. Search can be triggered automatically or selected directly, allowing the assistant to retrieve current information and provide links to relevant sources. It is designed to combine conversational follow-up with web retrieval and has expanded ChatGPT’s role into news, current events, shopping, travel, and other information-seeking workflows.

• Deep ResearchResearch agent2025

  An agentic research mode that plans and executes multi-step searches across the public web, uploaded files, and selected connected applications. It can propose a research plan, allow user adjustments, show progress, and produce a structured report with citations or source links. Deep Research is intended for complex questions requiring synthesis rather than quick factual lookup.

• ChatGPT agentComputer-use AI agent2025

  An agentic capability combining research, browsing, and computer interaction. It can perform multi-step tasks in a virtual computer environment, interact with websites, gather information, and execute actions under user direction. It evolved from OpenAI’s Operator preview and is designed to let users interrupt, redirect, or review tasks while they are running.

• Custom GPTs and GPT StoreConfigurable AI assistants and marketplace2023

  A product layer allowing users and organizations to create specialized versions of ChatGPT with custom instructions, knowledge, tools, and optional external actions. The GPT Store provides discovery, featuring, and distribution for eligible public GPTs, while business workspaces can manage internal GPTs and their availability. The system extends ChatGPT from a single assistant into a platform for task-specific experiences.

• ChatGPT ImagesAI image generation2023

  Image-generation functionality embedded in ChatGPT. Users can describe an image conversationally, ask for revisions, and in supported experiences provide existing images as references. OpenAI has emphasized improvements in text rendering, iterative editing, and integration with the surrounding conversation. Generated images may include C2PA provenance metadata indicating AI origin.

• ChatGPT HealthHealth and wellness AI experience2026

  A dedicated ChatGPT experience for health and wellness conversations. OpenAI describes additional protections including purpose-built encryption and isolation, with the ability in supported contexts to connect health records or wellness applications. It is intended to help users prepare for and understand health-related questions, not to replace clinicians or provide unrestricted professional medical care.

• 2023Icelandic language collaboration

  Iceland

  OpenAI worked with Icelandic volunteers and public figures to improve ChatGPT’s ability to converse in Icelandic. The collaboration was connected to broader efforts to preserve and strengthen use of the language in digital systems.

  Outcome. Improved attention to Icelandic-language support and demonstrated ChatGPT’s use in language preservation initiatives.

  Source
• 2023Albanian EU-document translation initiative

  Albania · European Union

  The Albanian government decided to use ChatGPT for rapid translation of European Union documents and for analysis of changes associated with the country’s accession process.

  Outcome. Positioned ChatGPT as an administrative translation and document-analysis tool, while also illustrating the need for official review of AI-assisted government work.

  Source
• 2022ChatGPT Feedback Contest

  United States · Global online audience

  OpenAI used a feedback-oriented contest and research-preview framing to encourage users to experiment with ChatGPT and report how the system performed across practical tasks. The initiative reflected the product’s early positioning as an openly accessible experiment whose value and risks would be learned through public use.

  Outcome. Helped generate early user feedback and public awareness during the research-preview period.

  Source

• 2026Launch ChatGPT HealthProduct launch

  Health questions were already a common use case, but general-purpose chats present heightened privacy, safety, and data-isolation requirements.

  What changed. OpenAI introduced a dedicated Health experience with additional privacy and compartmentalization measures and began a U.S. rollout to adults on web and iOS.

  Aftermath. ChatGPT gained a specialized health interface while OpenAI emphasized that the product should support informed conversations rather than replace professional medical care.

  Source
• 2026Test advertising in the free experienceStrategy

  Maintaining broad free access requires substantial computing and infrastructure expenditure. OpenAI pursued advertising as an additional monetization channel while attempting to keep ads distinct from generated answers.

  What changed. OpenAI began testing ChatGPT ads for eligible adult users and expanded the pilot to the United Kingdom, Mexico, Brazil, Japan, and South Korea. It also introduced additional ad-buying and measurement tools.

  Aftermath. Advertising became part of ChatGPT’s commercial experimentation, accompanied by restrictions on sensitive contexts, minors, vulnerable conversations, and misleading or unsafe ad categories.

  Source
• 2025Shift ChatGPT toward agentic task completionStrategy

  Answer generation alone captures only part of the value of an assistant. OpenAI pursued systems that could research, browse, operate software, and complete longer sequences of work under user supervision.

  What changed. OpenAI introduced Operator, Deep Research, Codex-related workflows, and ChatGPT agent capabilities, progressively integrating browsing, research, coding, and computer-use functions.

  Aftermath. The product proposition expanded from conversational assistance toward supervised automation, creating new utility as well as new safety, privacy, reliability, and authorization challenges.

  Source
• 2025Withdraw a sycophantic ChatGPT updateOther

  Users reported that a model update was overly agreeable and flattering, including in situations where a more cautious or challenging response was appropriate.

  What changed. OpenAI rolled back the update and stated that it was working to improve model behavior and safety around mental-health emergencies and emotionally dependent use.

  Aftermath. The incident demonstrated the difficulty of optimizing conversational warmth without encouraging harmful validation and became a prominent example of the risks of behavior changes in a widely used assistant.

  Source
• 2024Deploy GPT-4o and expand multimodal accessGeneration change

  OpenAI wanted to make advanced interaction faster and more natural while reducing the separation between text, vision, and audio experiences.

  What changed. OpenAI launched GPT-4o and rolled selected capabilities into the free ChatGPT tier, including expanded voice, vision, memory, and tool access subject to limits.

  Aftermath. ChatGPT became more explicitly multimodal and more accessible to free users, increasing its use for voice conversations, image understanding, and desktop workflows.

  Source
• 2024Add web search to ChatGPTStrategy

  The product’s static model knowledge limited its ability to answer questions about current events, prices, sports, news, and other changing information.

  What changed. OpenAI introduced ChatGPT Search, combining conversational context with web retrieval, third-party search providers, partner content, and links to sources.

  Aftermath. ChatGPT moved into closer competition with search engines and became more useful for current-information tasks, while raising new questions about publisher traffic, attribution, and content licensing.

  Source
• 2023Introduce ChatGPT PlusPrice change

  The initial research preview was free, but demand and capacity constraints created pressure for a paid access tier. OpenAI used a subscription model to provide more reliable access and support continued product development.

  What changed. OpenAI launched ChatGPT Plus at US$20 per month, with benefits including priority availability and access to newer capabilities. The price and included features have remained subject to later plan changes.

  Aftermath. The freemium structure became a foundation of ChatGPT’s commercial model, separating broad free access from paid capacity and feature entitlements.

  Monthly subscription price. Free research preview → US$20 per month for ChatGPT Plus at launch (February 2023 launch; pricing page current context accessed in 2026)

  Source
• 2023Launch GPTs and move toward a platform ecosystemProduct launch

  A single general-purpose assistant could not address every specialized workflow. OpenAI sought to let users and organizations customize ChatGPT without training a model from scratch.

  What changed. OpenAI introduced GPTs, which combine custom instructions, knowledge, and capabilities for specific purposes, followed by the GPT Store for discovery and sharing.

  Aftermath. ChatGPT expanded from a single assistant into a platform supporting user-created and organization-specific assistants.

  Source

• 2025Sycophancy and mental-health safety concernsControversy

  OpenAI withdrew a ChatGPT update after users and observers reported that it had become excessively flattering and too willing to affirm questionable beliefs or impulsive behavior. The episode intensified concern about emotional reliance, delusion reinforcement, and the safety of conversational systems used during mental-health crises.

  Source
• 2025Academic dishonesty and malicious-use criticismControversy

  ChatGPT has been criticized for enabling students to submit AI-generated work, producing misinformation, and assisting with malicious or harmful code. Schools, employers, and regulators have debated detection, disclosure, acceptable-use rules, and the limits of automated safeguards.

  Source
• 2024Copyright and training-data controversyControversy

  The use of copyrighted books, journalism, websites, and other material in training large language models has generated legal and ethical controversy. Authors, publishers, and other rights holders have argued that model training and output behavior may infringe their rights or undermine compensation and attribution systems.

  Source
• 2023Conversation-history and account-data exposure bugControversy

  A March 2023 software incident allowed some users to see titles of other users’ conversations and was later reported by OpenAI to have exposed additional account and payment-related metadata for a limited period. OpenAI temporarily restricted access while investigating and fixing the issue.

  Source
• 2023Hallucinations, misinformation, and reliability concernsControversy

  ChatGPT has been repeatedly criticized for producing confident-sounding but false or nonsensical answers, fabricated citations, and invented text. The problem became one of the defining limitations of the product and a central reason OpenAI and users emphasize verification, especially for high-stakes use.

  Source

• 2026ChatGPT Health launches in the United States

  OpenAI introduced a dedicated health and wellness experience with additional privacy and data-isolation controls; the U.S. rollout began for adults on web and iOS in July 2026.

  Product launchRegulation
  Source
• 2026ChatGPT advertising pilot expands internationally

  OpenAI expanded its ChatGPT advertising pilot from the United States to the United Kingdom, Mexico, Brazil, Japan, and South Korea while stating that ads would be kept separate from model answers and excluded from sensitive contexts.

  CampaignPricingOther
  Source
• 2026OpenAI expands ways to purchase ChatGPT advertising

  OpenAI introduced additional buying and measurement options for ChatGPT ads, including partner-supported purchasing, a beta self-serve Ads Manager, and cost-per-click bidding.

  Pricing
  Source
• 2025Operator introduced as a browser-using agent

  OpenAI released Operator as a research preview capable of interacting with websites through its own browser. In July 2025, the capability was integrated into ChatGPT agent mode.

  Product launch
  Source
• 2025Deep Research becomes a ChatGPT capability

  Deep Research added an agentic workflow that plans searches, reviews online and uploaded sources, and synthesizes findings into a cited report.

  Product launch
  Source
• 2024GPT Store begins rollout

  The GPT Store launched as a discovery and distribution marketplace for custom versions of ChatGPT created by users, partners, and organizations.

  Product launch
  Source
• 2024GPT-4o and expanded tools reach free users

  OpenAI launched GPT-4o and expanded access to voice, vision, image-related capabilities, memory, the GPT Store, and desktop workflows for free users, subject to limits.

  Product generationProduct launch
  Source
• 2024ChatGPT Search launches

  ChatGPT Search introduced web retrieval with links to relevant sources, allowing the assistant to answer more current questions and compete more directly with conventional search engines.

  Product launch
  Source
• 2023ChatGPT Plus subscription launches

  OpenAI introduced ChatGPT Plus as a paid subscription with more reliable access, faster responses, and priority access to new capabilities.

  PricingProduct launch
  Source
• 2023GPT-4 becomes available in ChatGPT Plus

  The GPT-4 model was added to ChatGPT Plus, materially expanding the service’s reasoning, writing, coding, and multimodal capabilities compared with the launch-era model.

  Product generation
  Source
• 2023ChatGPT iOS application launches

  OpenAI introduced an official iOS application with synchronized conversation history and voice input, beginning a broader expansion from the web into mobile computing.

  Product launch
  Source
• 2023GPTs customization system introduced

  OpenAI introduced GPTs, allowing users to create task-specific versions of ChatGPT using custom instructions, additional knowledge, and selected capabilities.

  Product launch
  Source
• 2022OpenAI publicly releases ChatGPT as a research preview

  OpenAI launched ChatGPT on November 30, 2022, initially presenting it as a free research preview intended to gather user feedback and identify the system’s strengths and weaknesses.

  Product launch
  Source

1. Introducing ChatGPT
2. ChatGPT Pricing
3. What is ChatGPT: FAQ
4. ChatGPT Capabilities Overview
5. Introducing the ChatGPT app for iOS
6. Introducing GPTs
7. Introducing the GPT Store
8. Memory and new controls for ChatGPT
9. Introducing GPT-4o and more tools to ChatGPT free users
10. Introducing ChatGPT search
11. Introducing Operator
12. Deep research in ChatGPT
13. Introducing ChatGPT Health
14. Testing ads in ChatGPT
15. New ways to buy ChatGPT ads
16. How People Use ChatGPT
17. ChatGPT
18. ChatGPT Feedback Contest