OpenAI & ChatGPT Timeline: GPT Release Dates to GPT-6.1 (2026)

ChatGPT launched on November 30, 2022, as OpenAI’s public AI assistant. This timeline tracks the GPT models and product releases that shaped ChatGPT, from GPT-1 and GPT-3 to GPT-4, GPT-5, GPT-5.6, GPT-6, Codex, and Sora.

The quick release table below lists the main model and product dates. The year-by-year timeline adds launches, access changes, APIs, subscriptions, safety updates, and other events from 2018 to the present.

GPT-6 Pro received a Chat update on October 16 with faster responses, improved tool use, and parallel agents for eligible Pro users.

GPT-6 with Intelligent UI began rolling out in ChatGPT on October 7, 2026. It adds interactive responses to the Chat tab, with Free and Go access scheduled to expand from October 8.

OpenAI’s latest Sol model for Work, Codex, and the API is GPT-6.1 Sol, announced at DevDay on September 29, 2026. The release improves coding, computer use, professional work, and factual accuracy over GPT-6 Sol.

GPT-6 Astra is OpenAI’s most capable model across the family. GPT-6.1 Sol approaches Astra on several evaluations at lower cost, while GPT-6 Luna is the fastest and least expensive GPT-6 model. GPT-6.1 Sol Ultrafast became available to all API users on October 8, 2026, with separate pricing and rate limits.

Last updated: October 22, 2026

GPT Model and ChatGPT Release Dates

Latest ChatGPT and OpenAI Releases

Decisions API general availability

The Decisions API became generally available on October 22, 2026. It returns probabilities, choices, and scores from text, images, and JSON, with up to 400 questions per request. The release adds native GPT-Live integration, prompt caching, and U.S. and EU inference residency. Input costs $0.10 per million tokens; cached reads cost $0.01. Cache writes have separate rates, and there are no output-token charges. Source

OpenAI models announced for Google Cloud

OpenAI announced on October 21, 2026, that its frontier models and Codex workflows are coming to Google Cloud Model Garden. Purchases will run through Google Cloud Marketplace and can draw down Google Cloud spending commitments. General availability is planned in the coming weeks. OpenAI models are also planned for Google Antigravity and Gemini Agent. Source

Locked chats

Locked chats began rolling out to Free, Go, Plus, and Pro users on October 20, 2026. They hide selected conversations from history, pinned chats, and search. Access requires a four-digit PIN or supported device authentication; retention and model-improvement settings remain unchanged. Source

Orchestra Bio acquisition

On October 20, 2026, OpenAI announced its acquisition of Orchestra Bio. Its team and drug-development platform will join Rosalind to help coordinate studies, decisions, and related work across drug programs. Source

ChatGPT Slides

ChatGPT Slides began rolling out on October 16 to eligible Pro, Business, and Enterprise users with Space access. Teams can edit decks together, comment, present, and export to PowerPoint, PDF, or Google Slides on the web. Source

GPT-6 Pro update

The October 16 GPT-6 Pro update adds faster Chat responses, better plugin and tool use, and parallel agents for eligible Pro users. Managed workspace access depends on settings and rollout eligibility. Source

Rosalind Academic Plan Opens Applications

On October 15, 2026, OpenAI introduced the Rosalind Academic Plan for eligible U.S. university labs. Approved labs can receive up to 30 ChatGPT Business Standard seats free for up to 12 months. GPT-Rosalind and Rosalind Workbench support life sciences research, subject to trusted-access approval and usage limits.

Principal investigators or lab leads apply for their labs. Universities with an OpenAI Enterprise agreement use their institutional access process. API access is excluded. Research data is not used for model training by default.

Access requires identity, institutional, and research-use checks, a designated safety contact, and ongoing monitoring. Approval must be followed by workspace setup and member activation.

ChatGPT Adds Chrome Tab Access and Computer Use

On October 15, 2026, OpenAI announced Chrome tab access and computer use for ChatGPT. In Chat, ChatGPT can read across open Chrome tabs and gather information for answers.

In Work, ChatGPT can control Chrome tabs to send messages, fill out forms, and download files. It can also organize tabs and place them into groups.

GPT-Live-1.1 and Quip Roll Out in ChatGPT Voice

On October 15, 2026, OpenAI began rolling out GPT-Live-1.1 and GPT-Live-1.1 Quip in ChatGPT Voice. Quip is the mini version of GPT-Live-1.1. The update improves accents across languages and adds more varied phrases during web searches.

New voices include French, German, Hindi, Japanese, Korean, Mexican Spanish, and Bubbie in English. Availability varies by region and account. Choose a voice in Settings > Voice.

GPT-Live-1.1 also brings the voice update to Codex and dots. The rollout began for all users on October 15. API access is coming soon.

AI for Educators Adds Brazilian Portuguese

OpenAI launched AI for Educators in Brazilian Portuguese on October 15, 2026. The free OpenAI Academy course helps primary, secondary, and higher-education teachers plan lessons, activities, assessments, and communications. Study Mode guides learning through questions, hints, and understanding checks.

ChatGPT Sites Removes Fixed Creation Limits

On October 15, 2026, OpenAI removed the fixed limit on Sites creation for Plus and Pro users. Usage and storage limits still apply.

GPT-6.1 Sol Ultrafast Opens to All API Users

On October 8, 2026, OpenAI added Ultrafast mode for GPT-6.1 Sol to the Responses API. The service tier reduces the time between generated output tokens and is available to all API users, subject to separate rate limits. Developers select gpt-6.1-sol with service_tier: "ultrafast".

For prompts with up to 272,000 input tokens, GPT-6.1 Sol Ultrafast costs $12 per million input tokens, $0.60 for cached input, $15 for cache writes, and $60 for output. Above that threshold, prices are $24 for input, $1.20 for cached input, $30 for cache writes, and $90 for output. Regional processing adds a 10% surcharge.

Default GPT-6.1 Sol Ultrafast limits are 1 million tokens per minute for Build, 4 million for Launch, and 40 million for Grow. GPT-6 Astra also supports Ultrafast for all API users, with default limits of 500,000, 1 million, and 5 million tokens per minute, respectively.

GPT-6.1 Sol Ultrafast supports global processing and U.S. and EU data residency. Astra Ultrafast supports global processing and U.S. data residency. Both support HTTP requests and persistent WebSocket connections. OpenAI recommends WebSockets for agents with frequent tool calls because network overhead can reduce the speed gains.

Codex Responds Sooner to Follow-Up Instructions

On October 8, 2026, OpenAI began rolling out faster steering in Codex in the ChatGPT desktop app. Codex can incorporate corrections, added information, and changes in direction sooner during a running task. Settings > General > Follow-up behavior controls whether follow-up messages steer the current run or wait for the next one.

OpenAI Reports Disruption of Two Covert Influence Operations

OpenAI published its false-front influence operations report on October 8, 2026. It banned ChatGPT account clusters linked to a Russia-origin operation, Dark Clark, and an Iran-origin operation, Bogus Bylines. The operators used AI for internal reports, deceptive content, and article pitches. The report describes a purported research organization in Latin America and seven fake journalist personas that placed articles in online outlets.

GPT-6 Brings Intelligent UI to ChatGPT

OpenAI introduced GPT-6 with Intelligent UI on October 7, 2026. Responses can combine text, visuals, charts, buttons, forms, and interactive tools such as calculators or games. ChatGPT can begin answering while it continues to reason or use tools.

The global rollout began October 7 for Plus, Pro, Business, and Enterprise and expands to Free and Go from October 8. Enterprise access depends on administrator settings. Paid tiers use GPT-6 Sol, and Free and Go use GPT-6 Luna, both tuned for everyday conversation.

Intelligent UI supports Instant through Extra High reasoning. The Pro reasoning option uses GPT-6 Astra without Intelligent UI. This release updates Chat; Work and Codex keep their existing models.

OpenAI reports that GPT-6 Instant starts web-search answers 44% sooner on average than GPT-5.6 Instant in its internal evaluation.

ChatGPT for Teens Adds Study Tools and Announces College Planner

OpenAI announced College Planner and new teen learning tools on October 7, 2026. College Planner is coming soon, initially for U.S. students in grades 10–12 who plan to attend a four-year college. It will organize application requirements, deadlines, tasks, and financial-aid steps.

The learning update adds saved flashcard decks, scheduled practice, and interactive quizzes from uploaded notes. Multi-page camera capture combines notes into one PDF on iOS; Android support is planned.

OpenAI also announced three years of support for Boston Children’s Hospital’s Digital Wellness Lab and its independent Student Advisory Council, plus support for College Advising Corps.

GPT-6 Chat Safety Update and chat-latest API Refresh

OpenAI published an October system-card update for the Chat versions of GPT-6 Sol and Luna. Both are treated as High capability in cybersecurity and biological and chemical domains, below Critical, and neither reaches the High threshold in AI self-improvement. Evaluations found stronger jailbreak resistance and lower dishonesty, deception, and guardrail circumvention than their GPT-5.6 Chat predecessors, alongside some lower-severity safety regressions.

OpenAI updated the chat-latest API snapshot on October 7, 2026, to reflect the latest ChatGPT model for Plus, Pro, Business, and Enterprise. The snapshot changes regularly; OpenAI recommends the GPT-6 model family for production API use.

OpenAI Publishes New AI Mathematics Results

On October 6, 2026, OpenAI published mathematical results from an unreleased internal frontier model. The GitHub release includes Lean proof formalizations, revision and citation protocols, 10 reasoning summaries, compute estimates, and attempted-problem statistics. The average result used compute equivalent to about three hours of ChatGPT Pro thinking. OpenAI is working toward releasing the model.

Decisions API Enters Public Beta

Released on October 6, 2026, the Decisions API uses GPT-6 Luna to return probabilities, fixed choices, or rubric scores from text and images. OpenAI reports roughly 10 times faster answers than the Responses API. The API became generally available on October 22; current pricing includes separate cached-read and cache-write rates.

ChatGPT Adds Audio File Uploads

On October 6, 2026, ChatGPT added audio file uploads for transcripts, recording summaries, and questions about recorded content. Access requires a paid subscription or workspace and varies by workspace settings, region, client version, and model. Transcripts can contain errors, and performance varies across languages.

GPT-6 Astra Advances on Ironclad Contracting Tasks

OpenAI announced its Ironclad research collaboration on October 6, 2026. Across 11 contracting tasks, GPT-6 Astra at Max reasoning scored 55.0%, versus 41.6% for GPT-5.6 Sol at High. Estimated time per attempt fell from 37.0 to 19.2 minutes. These are research results with simulated timing, not measured customer savings. These tasks require human oversight.

Atlassian Expands Its OpenAI Partnership

On October 6, 2026, Atlassian and OpenAI expanded their partnership to bring GPT-6 models to Atlassian’s platform and Rovo. Atlassian plugins connect ChatGPT and Codex to project information and documentation under existing permissions. Atlassian and OpenAI are exploring deeper Jira integrations for assigning agent work, tracking progress, and reviewing results.

API Usage Tiers and HIPAA Setup Change

On October 6, 2026, OpenAI reduced API usage tiers from five to three: Build, Launch, and Grow. Organizations upgrade automatically as total credit purchases reach tier thresholds. On October 5, eligible organization administrators gained an in-product flow to accept the standard Business Associate Agreement and enable HIPAA compliance support.

OpenAI Announces EU Text Watermarking

On October 5, 2026, OpenAI introduced textGrain, an invisible statistical watermark for text. API customers worldwide can opt in for select models; watermarking is off by default. Eligible ChatGPT and Codex text output across all plans in the EU will receive watermarks over the coming weeks.

The detector is initially restricted to approved researchers and expert organizations. Short passages, constrained text, editing, and translation can reduce detection reliability. A watermark does not identify a user, establish ownership, or verify accuracy; an undetected watermark does not prove human authorship.

ChatGPT Ads Adds a Visual Format and Measurement Tools

OpenAI announced a visual ad format on October 5, 2026. Testing will begin later in October during image generation in the U.S. with an initial group of advertisers. Ads will be labeled and separate from the generated image. Advertising does not influence ChatGPT’s answers.

The update expands conversion-data integrations, attribution partnerships, and early incrementality measurement. Qualifying advertisers can use Negative Phrases for placement restrictions. OpenAI is developing brand suitability evaluation pilots with DoubleVerify and Integral Ad Science that do not access private user conversations.

Finances, Shopping, and Document Scanning Expand

The October ChatGPT updates add Finances access for Free and Go users in the U.S. from October 2 on web, iOS, and Android. Connected financial accounts support questions about spending, savings, and investments.

On October 1, ChatGPT added virtual try-on for clothing and accessories through ChatGPT Images, plus product Favorites and folders on mobile and web. Reference photos are saved for future try-ons and can be changed or deleted in Settings. An iOS camera update also began rolling out Scan mode, which combines multiple pages into one PDF.

GPT-6.1 Sol Arrives at DevDay 2026

OpenAI introduced GPT-6.1 Sol on September 29, 2026. The model improves on GPT-6 Sol in agentic coding, complex documents, business workflows, computer use, scientific tasks, and factual accuracy. OpenAI says it matches GPT-6 Astra on DeepSWE 1.1 at roughly one-fifth of the cost per task, although Astra scored higher on the most difficult scientific tasks in OpenAI’s reported tests.

GPT-6.1 Sol is available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users, and in the API as gpt-6.1-sol. It is not yet available in standard Chat. API prices are $2 per million input tokens, $0.10 per million cached input tokens, and $10 per million output tokens. OpenAI reports improved performance on its alignment tests. GPT-6.1 Sol Ultrafast followed in the API on October 8, 2026, with separate pricing and rate limits.

Dots and Space Expand ChatGPT Work

OpenAI introduced dots, always-on agents powered by GPT-6 Astra that can work on a cloud computer, use connected apps, and carry out ongoing tasks. Users can inspect their work, set rules for actions, and review decisions that require approval. A first dot is included with Pro and Business Premium plans in eligible markets; Enterprise, Edu, and Healthcare workspaces can try a beta when administrators enable it.

ChatGPT Space replaces Library for Pro, Business, and Enterprise users. It brings shared pages, files, and team context into one workspace where people can collaborate with ChatGPT, Codex, and dots. Users can create and edit pages on web and desktop, while mobile access initially supports finding, reading, and sharing. Collaborative slides and spreadsheets are planned for a later release.

DevDay Adds Developer Tools and New Plan Options

At DevDay 2026, OpenAI announced GPT-6 Astra Ultrafast for eligible API users and Pro 500 and Enterprise users in ChatGPT Work and Codex. The company also added computer use to the Agents API, previewed a Decisions API for predefined choices, expanded cloud Codex and Code Review, and introduced Codex Security Cloud for eligible paid plans. GPT-6.1 Sol Ultrafast followed in the API on October 8, 2026.

ChatGPT gained plugin extensions with sidebar panels, improvements to plugin creation and discovery, and support for plugins on Sites in eligible workspaces. Business and Enterprise plans gained team sharing and ChatGPT access from Slack and Microsoft Teams. Sign in with ChatGPT lets Plus and Pro users use eligible plan allowances in participating apps, subject to their existing usage limits. The new Pro 500 tier costs $500 per month and includes Astra Ultrafast. OpenAI also opened interest in its Marketplace, where eligible enterprise customers can apply part of an existing OpenAI commitment to approved partner products.

GPT-6 Sol and Luna Extend the GPT-6 Model Family

OpenAI introduced GPT-6 Sol and GPT-6 Luna on September 22, 2026. Both models use training methods related to GPT-6 Astra and improve professional work, factual reliability, coding, computer use, and alignment at lower cost.

GPT-6 Sol and Luna are available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users. Free and Go users can access GPT-6 Luna in the desktop app. At the September launch, access was limited to Work, Codex, and the API. Everyday-conversation versions began rolling out in Chat on October 7. API developers can use gpt-6-sol and gpt-6-luna.

API pricing per 1 million tokens is $2 for input and $10 for output with GPT-6 Sol, and $0.10 for input and $0.50 for output with GPT-6 Luna. OpenAI describes both prices as 50% lower than the promotional prices of their GPT-5.6 predecessors. Cached input-token reads receive a 90% discount.

OpenAI reports that GPT-6 Sol made about half as many mistakes as GPT-5.6 Sol in an internal factuality evaluation. GPT-6 Sol scored 68.8% on DeepSWE 1.1 at maximum reasoning effort, while GPT-6 Luna scored 66.6%. OpenAI classifies both models as High capability in the Cybersecurity and Biological and Chemical domains, below the Critical level reached by Astra in cybersecurity.

Agents API Brings the Codex Harness to Developers

OpenAI introduced the Agents API in public beta on September 10, 2026. Developers can use it to build and run cloud agents with the Codex harness that manages context, tools, long-running sessions, and optional subagents.

The API supports OpenAI-hosted sandboxes, customer-managed infrastructure, and integrated sandbox providers. Agents can use MCP servers, custom functions, web search, tool search, programmatic tool calling, and context compaction. The public beta has no separate platform fee; developers pay for the tokens and tools their agents use.

GPT-Live-1 Adds Full-Duplex Voice to the API

OpenAI released GPT-Live-1 in the API on September 10, 2026. The model listens and speaks at the same time, so voice agents can handle interruptions, pauses, background noise, and changes of direction without relying on a chained speech-to-text, language-model, and text-to-speech architecture.

Developers can set the agent’s tone, pace, and style, use turn detection and transcripts, and delegate deeper reasoning or tool calls to models such as GPT-6 Astra. GPT-Live-1 is available at $0.05 per minute for the front-end voice layer, with the selected backend model and agent tools billed separately.

New ChatGPT Work Products for Data and Financial Analysis

OpenAI introduced a Data agent for ChatGPT Work on September 10, 2026. The plugin connects to approved company data sources and business definitions, investigates questions, creates interactive dashboards, and can share findings through connected tools after user approval. Administrators choose the available connections and user roles, while queries enforce the source account’s existing permissions.

On the same day, OpenAI announced ChatGPT for Financial Services, a tailored ChatGPT Work experience for eligible financial institutions. It combines GPT-6 Astra with built-in financial data, connected sources, firm templates, and enterprise security controls for research, financial models, and client materials.

Codex and ChatGPT Support Antimicrobial Research

OpenAI also described how César de la Fuente’s laboratory uses Codex and ChatGPT to explore genome and protein datasets for potential antimicrobial molecules. Lab members use the tools to develop and refine code, process datasets, analyze results, and form hypotheses across biology, chemistry, and computer science. The researchers emphasize that laboratory experiments is essential for validating AI-generated candidates.

ChatGPT Images 2.5 Adds Faster Generation and More Editing Control

OpenAI introduced ChatGPT Images 2.5 on September 8, 2026. The new image model is rolling out to ChatGPT, ChatGPT Work, and Codex users across all plans on desktop, mobile, and web.

Images 2.5 produces sharper details, more natural lighting and textures, and more recognizable subjects from reference photos. It follows focused editing instructions more reliably across multiple turns and reduces image-generation latency by up to 50% compared with Images 2.0.

ChatGPT also adds Sketch for drawing a visual reference, templates for formats such as posters and merchandise, comments placed directly on images, and optional prompt sharing. For API users, OpenAI released GPT-Image-2.5 Flare for most applications and GPT-Image-2.5 Sunburst for workflows that need tighter editing control. The release retains prompt and image safety checks, C2PA metadata, and invisible watermarking.

GPT-6 Astra Is Released with Critical Cybersecurity Capability

OpenAI released GPT-6 Astra on September 3, 2026. The rollout started with a limited set of organizations and is expanding to ChatGPT Plus, Pro, Business, and Enterprise users, the OpenAI API, and AWS. Enterprise administrators must enable Astra, and access is off by default at launch. Users on Pro, Business, and Enterprise plans also receive access to GPT-6 Astra Pro.

Astra is designed for computer use, browsing, software engineering, cybersecurity, science, and professional work. OpenAI reports scores of 98% on FrontierMath Tier 4, 99.9% on ARC-AGI-3, and 100% on ExploitBench. The model can also fill online forms, update records, organize calendars, research information, and work with documents, spreadsheets, and code.

GPT-6 Astra is OpenAI’s first model to reach the Critical cybersecurity capability threshold under its Preparedness Framework. In testing without production safeguards, it found and used two previously unknown zero-day vulnerabilities and achieved a 100% ExploitBench score. The deployed model refuses more advanced cyber tasks such as creating proof-of-concept exploits; OpenAI says Daybreak will expand access to less restrictive safeguards for approved defensive work.

OpenAI’s safety overview reports that Astra is more robust to jailbreaks and prompt injection than GPT-5.6 Sol and produced roughly half as many higher-severity misalignment flags in a simulation of more than 54,000 internal Codex tasks. The company has added misalignment monitoring to all tool-using external Astra inference. OpenAI also found that Astra can be harder to monitor in adversarial tests, so the system card treats monitorability as an ongoing safety issue.

Daybreak for Frontline Defenders Commits $1 Billion to Cyber Defense

OpenAI introduced Daybreak for Frontline Defenders on September 3, 2026, with a $1 billion commitment for subsidized access to Daybreak cyber models and products, training, technical support, and partnerships. The funding is intended to be used over the next six months, starting with resource-constrained defenders in the United States and expanding to partner countries.

The initiative prioritizes water and wastewater systems, electric grid operators, state and local governments, community and regional banks, nonprofits, open-source maintainers, and similar organizations. Defenders can use Daybreak to review legacy code, analyze suspicious activity, identify and validate vulnerabilities, prioritize risks, and develop and test fixes.

OpenAI says thousands of defenders in 2,000 approved organizations and workspaces already use Daybreak. Its Daybreak Defense Network includes more than 35 enterprise products and partner-operated services, and a new MS-ISAC pilot will pair Daybreak access with training and hands-on support for public-sector and water-system defenders.

ChatGPT Connects to Epic and Public Healthcare Data

OpenAI added a new Epic integration and Healthcare Public Data plugin to ChatGPT for Healthcare on September 1, 2026. Healthcare organizations can bring authorized patient information from Epic into ChatGPT or integrate ChatGPT into supported electronic health record layouts.

The Healthcare Public Data plugin connects nine official sources, including ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed. Teams can work with specific records, fields, identifiers, and versions for clinical research, medication information, coverage policies, trials, and provider records.

Physicians evaluated the Epic-connected experience across 27 clinical use cases and rated 99.1% of 4,363 responses safe. ChatGPT for Healthcare customers can ask a workspace administrator to enable both connections. Eligible U.S. ChatGPT for Clinicians users can install the public-data plugin, but the Epic integration is not available to individual accounts.

ChatGPT Work and Scheduled Tasks Updates

OpenAI updated ChatGPT Work on August 25, 2026, with support for websites that require sign-in on web and mobile. The user completes the sign-in without ChatGPT seeing the username or password.

Scheduled tasks can now respond when something changes in Slack, Gmail, or GitHub for Plus and Pro users. OpenAI also began rolling out tasks to Free users, with a limit of three tasks, and added task sharing so other people can customize and use a shared task.

ChatGPT Business Premium Seats

ChatGPT Business now offers Premium seats with five times the usage of Standard seats and no five-hour usage limit. A Premium seat costs $100 per user per month with annual billing or $125 with monthly billing. Business workspaces can mix Standard and Premium seats and reassign them as team needs change.

ChatGPT for Teens

OpenAI introduced ChatGPT for Teens on August 18, 2026. It is designed to help teens learn, think critically, deepen understanding, and use AI with confidence. OpenAI says its system automatically places users it estimates are under 18, as well as users who state that they are between 13 and 17, into this experience.

The learning features include Study Mode, responsible homework reminders, Quizzes, Learning Visualizations, and Study Hours. Study Mode uses guiding questions and step-by-step support, while homework reminders can redirect apparent attempts to shortcut an assignment toward collaborative problem solving. Teens or parents can choose times when Study Mode is on by default.

Protections are enabled by default for teen users. The experience applies age-appropriate safeguards in higher-risk areas such as self-harm, violence, eating disorders, dangerous activities, and explicit sexual or graphic content. Parents with linked teen accounts can set Quiet Hours, manage selected settings, and receive limited safety notifications, including new notifications related to eating disorders.

ChatGPT for Teens also adds break reminders, clear cues that identify ChatGPT as AI, sensitive-image upload reminders, teen-specific onboarding, and simple customization options. OpenAI’s updated under-18 model spec says ChatGPT should not use romantic language, encourage emotional dependence, or imply that it has feelings or consciousness. OpenAI also announced a partnership with CodeAI to give students and educators more resources for learning how AI works and how to question, direct, and create with it.

GPT-5.6 Sol Ultrafast Mode Preview

OpenAI previewed Ultrafast mode for GPT-5.6 Sol on August 13, 2026. The preview can run GPT-5.6 Sol at up to 14 times the speed.

ChatGPT Computer History

Computer History turns activity across selected apps and websites into memories and a timeline that ChatGPT and Codex can reference later. You can return to recent work without reconstructing every open app, document, and next step.

Computer History is off by default in the ChatGPT desktop app on macOS. Pro users can turn it on. Business and Enterprise users need access from a workspace administrator before they can opt in. The feature requires Memories, does not work with an API key or Amazon Bedrock, and is unavailable in the European Economic Area, Switzerland, and the United Kingdom.

You control which apps and websites contribute. Computer History records interaction events and creates text summaries. It does not capture your screen or audio. Collection can be paused from the macOS menu bar, and each history item can be inspected or deleted.

GPT-5.6-Cyber and Expanded OpenAI Daybreak Access

OpenAI introduced GPT-5.6-Cyber on August 10, 2026. The cybersecurity-specific model is built on GPT-5.6 Sol and trained for advanced work such as finding zero-day vulnerabilities, developing exploit chains, validating exploits, and conducting authorized security tests.

The Daybreak program has two access tiers. Daybreak Blue gives approved defenders access to frontier general-purpose models, including GPT-5.6 Sol, with safeguards adjusted for secure code review, malware analysis, incident response, vulnerability discovery, and patch validation. Daybreak Red includes purpose-trained cyber models for higher-risk authorized research, including GPT-5.6-Cyber.

In OpenAI’s internal Advanced Cybersecurity Completion Rate evaluation, GPT-5.6-Cyber completed 95.0% of requests involving advanced scenarios such as exploit-chain development, authentication bypass, and privilege escalation. GPT-5.5-Cyber completed 57.3%, while GPT-5.6 Sol completed 1.5% with standard safeguards and 2.0% through Daybreak Blue.

The specialized model did not lead every evaluation. GPT-5.6-Cyber outperformed GPT-5.6 Sol and GPT-5.5-Cyber on ExploitGym and scored higher than GPT-5.6 Sol in OpenAI’s zero-day discovery evaluation. GPT-5.6 Sol produced stronger results in OpenAI’s vulnerability discovery and report-writing evaluation and performed best in the standard 300-turn ExploitBench setting.

OpenAI used GPT-5.6-Cyber to investigate Chrome’s V8 JavaScript engine and reported two previously unknown vulnerabilities to Google. Google fixed the disclosed issue as CVE-2026-15903. OpenAI also reported findings in a mobile operating system, a popular database, and an operating system kernel. Coordinated disclosure and remediation are underway.

OpenAI assessed GPT-5.6-Cyber as High for cybersecurity capability but below the Critical threshold. Access to both Daybreak tiers is limited to approved individuals and organizations conducting authorized work. Controls include identity verification, account security, monitoring, approved-use restrictions, and legal attestations. Individual Daybreak accounts must use hardware security keys beginning September 1, 2026.

GPT-5.6 Sol and Luna in ChatGPT

OpenAI updated GPT-5.6 Sol in ChatGPT on August 6, 2026, for Plus and Pro users. The model adjusts detail to each question, uses tighter formatting, and pushes back more often when a correction would help.

At the August launch, the same GPT-5.6 Sol model handled both Instant responses and deeper reasoning in ChatGPT. A slider on the web, mobile, and desktop apps let Plus and Pro users choose how much thought the model used for each response.

OpenAI says answers containing at least one factual error were about 68% less common with GPT-5.6 Sol than with GPT-5.5 Instant in an internal evaluation of financial, medical, and legal prompts. The corresponding reduction for GPT-5.6 Luna was about 62%.

OpenAI announced GPT-5.6 Luna as the default model for Free and Go users during the week of August 6. It also announced unlimited text chats and a Think button for the following week. Abuse guardrails apply, and file uploads, images, and other tools have separate limits.

The August version of GPT-5.6 Sol was tuned for everyday conversations in standard Chat. The October GPT-6 rollout replaces it with an everyday-conversation version of GPT-6 Sol.

The release also adds age-appropriate protections for users OpenAI believes are under 18. The model avoids romantic roleplay and age-restricted challenges, applies tighter boundaries to sensitive and dangerous content, and encourages contact with trusted people when a teenager may need support.

OpenAI Presence

OpenAI introduced OpenAI Presence on July 22, 2026. The enterprise product supports voice and chat agents that can answer questions, resolve issues, use company systems, take approved actions, and transfer work to people when needed.

Each deployment starts with a defined job and receives only the knowledge and system access required for that work. The customer sets permissions, policies, approval requirements, and escalation rules. Before launch, simulations and graders test outcomes, policy compliance, tool use, and escalation behavior.

After launch, production sessions and quality signals show where the agent needs attention. Codex uses the Presence plugin to investigate those signals and propose updates. Teams can compare each proposal with the production version, test it, and approve a controlled rollout.

Presence is available to eligible enterprise customers through a limited general availability program. OpenAI Forward Deployed Engineers and selected global systems integrators lead deployments, and the product is not available for self-service access. OpenAI says Presence now resolves 75% of inbound issues on its English-language phone support channel without human assistance. Its improvement process also reduced human handoffs by 15 percentage points in 10 days.

GPT-6

The GPT-6 family began with GPT-6 Astra on September 3, 2026. Astra is the highest-capability model in the family. GPT-6 Sol and GPT-6 Luna followed on September 22, and GPT-6.1 Sol arrived on September 29 with stronger coding, computer-use, and professional-work results than GPT-6 Sol.

Sol and Luna are available through ChatGPT Work, Codex, and the OpenAI API. GPT-6 Luna is also available to Free and Go users in the desktop app. Everyday-conversation versions began rolling out in standard Chat on October 7, with Sol for paid tiers and Luna for Free and Go from October 8.

The API model IDs are gpt-6-sol and gpt-6-luna. Sol costs $2 per million input tokens and $10 per million output tokens. Luna costs $0.10 per million input tokens and $0.50 per million output tokens. Luna also powers the Decisions API, which has separate input-only pricing.

GPT-5.6

OpenAI introduced GPT-5.6 Sol on June 27, 2026 in a limited preview of the GPT-5.6 series. The series has three tiers: Sol for the highest capability, Terra for everyday work, and Luna for faster, lower-cost use.

Sol is the top tier in the release. It adds a max reasoning effort for harder problems and an ultra mode that can use subagents for complex work.

On August 13, 2026, OpenAI previewed a separate Ultrafast mode that can run GPT-5.6 Sol at up to 14 times the speed.

Terra balances capability and cost. Luna targets cost-sensitive, high-volume workloads.

OpenAI first limited access to selected trusted partners through the API and Codex, then began a broader rollout across ChatGPT, Codex, and the API on July 9, 2026. The August 6 ChatGPT update gives Plus and Pro users the new chat-tuned version of Sol and expands Luna access for Free and Go users.

GPT-5.6 Sol improves results in coding, biology, and cybersecurity evaluations. It sets a new state of the art on Terminal-Bench 2.1, performs better than GPT-5.5 on GeneBench v1 with fewer tokens, and improves long-horizon security tasks in ExploitBench and ExploitGym.

Because GPT-5.6 is stronger in dual-use areas, the preview includes safeguards across the model, generation-time checks, account-level review, access controls, monitoring, enforcement, automated red teaming, and expert testing. GPT-5.6 Sol did not cross OpenAI’s Cyber Critical threshold in the reported tests.

OpenAI used more than 700,000 A100-equivalent GPU hours for automated red teaming aimed at universal jailbreaks. The tests targeted attacks designed to work across many prompts.

API pricing is calculated per 1 million tokens: Sol at $4 input and $20 output, Terra at $2.00 input and $12 output, and Luna at $0.2 input and $1.2 output. All three models have a 1,050,000-token context window and a maximum output of 128,000 tokens. Cached input costs $0.40 for Sol, $0.20 for Terra, and $0.02 for Luna.

AI LLM API Pricing: GPT, Gemini, Claude Fable/Mythos, and More

OpenAI and ChatGPT Timeline

2026

2025

2024

2023

2022

2020

2019

2018

GPT Model Generations Explained

GPT-1 used generative pre-training to create one model that could handle many language tasks.

GPT-2 made generated text much more visible. Its staged release also put model misuse and release policy into the public conversation.

GPT-3 turned the GPT line into a practical platform. Its few-shot ability made prompt-based writing, coding, summarization, and analysis useful to developers and early AI products.

ChatGPT gave the public a normal chat interface for GPT models. Questions, draft revisions, explanations, and code no longer required developer tools.

GPT-4 raised expectations for accuracy, reasoning, and multimodal input. It also became the model that many paid ChatGPT users associated with serious work.

GPT-4o moved ChatGPT closer to real-time multimodal use. Text, audio, vision, and later image generation became part of the same product direction.

The o-series models put more emphasis on step-by-step reasoning. o1-preview, o3-mini, o3, and o4-mini were built for harder math, science, coding, visual reasoning, and tool-use tasks.

GPT-5 brought earlier model directions into one generation for ChatGPT, the API, and Codex. GPT-5.6 extended that path with Sol, Terra, and Luna tiers. GPT-6 Astra moved the family into a new generation focused on computer use, browsing, software engineering, science, professional work, and cybersecurity. GPT-6 Sol and Luna added faster, less expensive options. GPT-6.1 Sol improved coding, computer use, scientific tasks, and factual accuracy while retaining Sol’s standard API token prices. The October Chat release added everyday-conversation versions of Sol and Luna with Intelligent UI. GPT-6.1 Sol gained the Ultrafast API service tier on October 8, with separate pricing and rate limits.

GPT-Rosalind is a specialized life sciences model with trusted access. The October 15 Rosalind Academic Plan expands sponsored access for eligible U.S. university labs. GPT-Live-1.1 began rolling out to all users across ChatGPT Voice, Codex, and dots that day, with API access coming soon. ChatGPT Voice also added the mini variant Quip.

Major ChatGPT Product Milestones

ChatGPT launched in November 2022 as a public research preview. The product was simple: a chat box, a model, and fast feedback from everyday users.

ChatGPT Plus arrived in February 2023 as the first paid consumer plan. Subscribers received better access during busy periods and earlier access to new features.

Plugins connected ChatGPT with outside tools and services. Later ChatGPT systems incorporated tool use and actions directly.

GPT-4o made multimodal ChatGPT feel more immediate. Voice, vision, faster responses, and later image generation became central parts of the product.

Canvas added a dedicated workspace for writing and coding inside ChatGPT. Drafts and code could be edited outside the message thread.

ChatGPT search brought current web information into the product, with source links for answers that depend on fresh information.

Dots introduced ongoing agent work through a cloud computer and connected apps, with activity review and action controls. ChatGPT Space gave teams shared pages and files where people, ChatGPT, Codex, and dots can work together.

Computer History added an optional source of personal context to the macOS desktop app. It can turn activity across selected apps and websites into memories and a timeline that help ChatGPT and Codex continue recent work.

ChatGPT for Teens added a learning-focused experience for teen users. Study Mode, responsible homework reminders, Quizzes, Learning Visualizations, and Study Hours support active learning, while default age-appropriate safeguards, parental controls, break reminders, and clear AI cues support safer, more balanced use.

ChatGPT for Healthcare connected authorized Epic records and official public healthcare datasets to a governed workspace. Clinical, research, and administrative teams can review patient context and structured information under existing permissions and enterprise controls.

Codex made software engineering a separate OpenAI product line. Developers can hand off coding tasks, inspect changes, and work across repositories.

Sora added video and audio generation to OpenAI’s product lineup. Sora 2 improved realism, physics, audio, and control.

ChatGPT Images made image generation and editing part of the ChatGPT workflow. GPT Image 1.5, ChatGPT Images 2.0, and ChatGPT Images 2.5 improved image quality, editing control, and generation speed. Images 2.5 also added Sketch, templates, direct image comments, and prompt sharing.

GPT-5.6 Luna extended the current model generation to Free and Go users. OpenAI announced Luna as their default model, with unlimited text chats subject to abuse guardrails and a Think button for questions that need more reasoning.

OpenAI Presence brought voice and chat agents into a managed enterprise deployment product. Companies define each agent’s job, system access, policies, approval requirements, and escalation rules. OpenAI and its deployment partners help test, launch, and improve the agent after it enters production.

OpenAI Daybreak grew from a cyber-defense access program into Daybreak for Frontline Defenders. OpenAI committed $1 billion in subsidized access, training, technical support, and partnerships for organizations protecting water, electricity, local government, banking, and other essential services.

Related Resources:

• Anthropic Claude Timeline
• Google AI & Gemini Timeline
• Meta AI Timeline
• Codex Timeline: Release Dates and Major Updates
• How to use DeepSeek V4 Pro or Flash in Codex and cut API costs
• DeepSeek Timeline: Model Release Dates and Key Milestones
• SpaceXAI Timeline: Grok Release History and Product Milestones
• Qwen Timeline: Model & Product Release History
• What Are The Rate Limits For OpenAI API?
• What Is The Max Token Limit In OpenAI’s GPT Models
• 10 Best ChatGPT Browser Extensions To Upgrade Your Workflow
• The Ultimate Codex Resource List: Skills, Plugins, MCP, SDKs & More

FAQs

Q: Are OpenAI models available on Google Cloud yet?
A: The October 21 announcement schedules general availability for the coming weeks.

Q: Can ChatGPT use my Chrome tabs?
A: Yes. Chat can read across open Chrome tabs to gather information. Work can control tabs to send messages, complete forms, download files, and organize tab groups.

Q: Is GPT-Live-1.1 available in the API?
A: API access is coming soon. The October 15, 2026 rollout brings the voice update to all users across ChatGPT Voice, Codex, and dots.

Q: Who qualifies for the Rosalind Academic Plan?
A: Eligible labs at U.S. universities can apply through their principal investigator or lab lead. Trusted-access approval is required. The offer excludes API access.

Q: What is GPT-Live-1.1 Quip?
A: Quip is the mini version of GPT-Live-1.1. Both began rolling out in ChatGPT Voice on October 15, 2026.

Q: Is GPT-6.1 Sol Ultrafast available?
A: Yes. It became available to all API users in the Responses API on October 8, 2026. Use gpt-6.1-sol with service_tier: "ultrafast". Separate pricing and rate limits apply.

Q: Does GPT-6.1 Sol Ultrafast support EU data residency?
A: Yes. It supports U.S. and EU data residency and global processing. GPT-6 Astra Ultrafast supports U.S. data residency and global processing. Regional processing carries a 10% surcharge.

Q: Can Free users use GPT-6 and Intelligent UI?
A: Free and Go rollout starts October 8, 2026, with GPT-6 Luna. Access expands gradually.

Q: Is College Planner available yet?
A: OpenAI announced it as coming soon on October 7, 2026. The initial experience is planned for U.S. students in grades 10–12 who intend to attend a four-year college.

Q: What is the Decisions API?
A: It is a GPT-6 Luna API for probabilities, choices, and scores. It became generally available on October 22, 2026.

Q: Has OpenAI released the model behind its new mathematics results?
A: No. OpenAI published research results from an internal frontier model on October 6, 2026, and is working toward releasing it.

Q: Does ChatGPT watermark text?
A: On October 5, 2026, OpenAI announced watermarking for eligible EU ChatGPT and Codex output over the coming weeks. API watermarking is optional for select models worldwide. The text detector initially requires approval.

Q: Will ads appear in ChatGPT-generated images?
A: OpenAI plans a U.S. visual ad test during image generation later in October 2026. Ads will appear separately from the generated image, carry a label, and leave ChatGPT’s answers independent.

Q: What is the latest GPT release in ChatGPT?
A: GPT-6 with Intelligent UI is the latest Chat release, announced on October 7, 2026. GPT-6.1 Sol, introduced September 29, is the latest Sol release for Work, Codex, and the API. GPT-6 Astra is the most capable model in the family.

Q: Where can I use GPT-6.1 Sol?
A: Plus, Pro, Business, Enterprise, and Edu users can access it in ChatGPT Work and Codex. Developers can use gpt-6.1-sol through the API. It is not yet available in standard Chat.

Q: What are dots and ChatGPT Space?
A: Dots are ongoing agents that use a cloud computer and connected apps to work on tasks under user controls. Space is a shared home for pages, files, and collaboration with people and AI. Availability depends on plan and workspace settings.

Q: What are GPT-6 Sol and GPT-6 Luna?
A: GPT-6 Sol and Luna are lower-cost GPT-6 models released on September 22, 2026. Sol is designed for demanding professional, coding, and computer-use tasks. Luna is the fastest and least expensive model in the GPT-6 family. Both are available in Work, Codex, and the API. Everyday-conversation versions began rolling out in Chat in October, with Sol for paid tiers and Luna for Free and Go.

Q: What is the Agents API?
A: The Agents API is OpenAI’s public-beta API for building and running cloud agents with the Codex harness. It supports managed or customer-selected sandboxes, tools, long-running sessions, and optional subagents. There is no separate platform fee; developers pay for tokens and tools used by their agents.

Q: What is GPT-Live-1?
A: GPT-Live-1 is a full-duplex voice model released in the API on September 10, 2026. It listens and speaks simultaneously, handles natural interruptions, and can delegate reasoning and tool calls to a backend model. The front-end voice layer costs $0.05 per minute.

Q: When was ChatGPT released?
A: ChatGPT was released on November 30, 2022, as a public research preview.

Q: Can Free users create scheduled tasks in ChatGPT?
A: Yes. OpenAI began rolling out scheduled tasks to Free users on August 25, 2026. Free users can create up to three tasks. Plus and Pro users can also create tasks that respond to changes in Slack, Gmail, and GitHub.

Q: What is ChatGPT for Teens?
A: ChatGPT for Teens is a learning-focused ChatGPT experience with Study Mode, responsible homework reminders, Quizzes, Learning Visualizations, and Study Hours. It also applies stronger default protections for users under 18, adds parent controls and healthy-use features, and automatically places users in the experience when OpenAI estimates they are under 18 or they state that they are 13 to 17.

Q: What is ChatGPT Images 2.5?
A: ChatGPT Images 2.5 is OpenAI’s image-generation and editing model released on September 8, 2026. It improves image detail, reference-photo fidelity, precise edits, and consistency across multiple edits, while reducing generation latency by up to 50% compared with Images 2.0. It is rolling out to ChatGPT, ChatGPT Work, and Codex users across all plans.

Q: Has OpenAI released Astra?
A: Yes. OpenAI released GPT-6 Astra on September 3, 2026. It began rolling out to a limited set of organizations and is expanding to ChatGPT Plus, Pro, Business, and Enterprise users, the OpenAI API, and AWS. Enterprise access is off by default at launch, and Pro, Business, and Enterprise plans also include GPT-6 Astra Pro.

Q: Can ChatGPT connect to Epic health records?
A: Yes, for supported organizational deployments. ChatGPT for Healthcare can use authorized patient context from Epic, and supported deployments can place ChatGPT inside an electronic health record layout. The Epic integration is not available to individual accounts.

Q: What is Daybreak for Frontline Defenders?
A: It is OpenAI’s $1 billion initiative to subsidize Daybreak cyber-model access, training, technical support, and partnerships for organizations that protect essential services. The program prioritizes resource-constrained defenders and includes a public-sector and water-focused pilot with MS-ISAC.

Q: Which model powers Free and Go ChatGPT accounts?
A: GPT-6 Luna begins replacing GPT-5.6 Luna in Chat from October 8, 2026. The rollout is gradual. File uploads, images, and other tools have separate limits.

Q: Does ChatGPT Computer History record the screen or audio?
A: No. Computer History records interaction events and creates text summaries. It does not capture the screen or audio. The feature is off by default, and users control which apps and websites contribute.

Q: What is the difference between GPT and ChatGPT?
A: GPT is OpenAI’s model family. ChatGPT is the assistant product that uses GPT and other OpenAI models through chat, search, files, voice, images, and related tools.