Short Answer

OpenAI is expected to release a personal AI agent satisfying market criteria before November 1, 2026, driven by anticipation for an announcement at DevDay on September 29, 2026, and a subsequent fast rollout to paid tiers, priced at 80%.

1. Market Behavior & Drivers

The market's probability for an OpenAI personal agent release before November 2026 surged from 3% to 54% in a single day. This repricing on September 24 was driven by reporting and leaks of a new OpenAI product, codenamed "o," described as an "always-on" personal AI assistant. Details from these reports indicated the rumored product would meet the market's specific resolution criteria for a general-purpose agent capable of multi-step tasks.
A subsequent 16-point spike to 80% on September 26 followed further reports about the same unconfirmed agent. These reports, reportedly originating from social media and tech communities, cited specific app and configuration references. Traders likely interpreted these details as tangible evidence of an impending product announcement, consolidating conviction around a near-term release. The rapid rally from near-zero to high probability was based entirely on this single, unconfirmed product narrative.
  • A release before November 1, 2026, is likely given the anticipated September 29 DevDay announcement.
  • A January 1, 2027, release appears highly plausible following a September 29 announcement.
  • A release before October 1, 2026, is unlikely due to the narrow two-day rollout window.

Who Wins and Why

Outcome Market Model Why
Before Oct 1, 2026 69.0% 64.0% A two-day window post-announcement is unlikely for a live, non-beta product release.
Before Nov 1, 2026 80.0% 75.0% An announcement at DevDay allows time for a compliant product roll out by November 2026.
Before Dec 1, 2026 74.0% 75.0% An announcement at DevDay provides sufficient time for a product rollout by December 2026.
Before Jan 1, 2027 78.0% 75.0% An earlier announcement makes a January 2027 release highly plausible for wider availability.

Current Context

Prediction markets forecast a personal agent release by November 2026. As of September 26, 2026, the Kalshi/CoinRithm market shows an approximately 80% probability for OpenAI to release a personal AI agent meeting contract specifications before November 1, 2026 [1][2]. The contract defines a general-purpose agent that remembers users, starts multi-step tasks across third-party apps, is live for U.S. users, and is not beta, test, preview, or employee-only access [1][2]. OpenAI initially released ChatGPT agent on July 17, 2025, for paid tiers, but current Help Center documentation indicates it is no longer a standalone product, directing users to ChatGPT Work for complex tasks [3][4][5]. This market date represents a resolution deadline and trader belief, not an official OpenAI roadmap or a reliable probability for a consumer-agent announcement [1][2][6][7][8].
OpenAI's product trajectory leans towards agentic functionality within ChatGPT. The company announced ChatGPT Work on July 9, 2026, describing it as an agent capable of acting across applications and files, sustaining projects for hours, and generating various finished documents; a phased rollout began with higher-tier paid plans [9][10]. Separately, workspace agents, introduced in April 2026 for Business, Enterprise, and Education users, focus on shared, long-running workflows within organizational permissions, distinguishing them from consumer-oriented personal agents [11]. Recent media signals, including Axios on September 20, 2026, and Forbes on September 21, 2026, point to personal assistants arriving and managed agents potentially featured at OpenAI's September 29 DevDay [12][13]. OfficeChai also reported on September 26, 2026, an unconfirmed always-on assistant code-named 'o' [14]. None of these sources establishes an official consumer launch date [14][12][13].
Consumer agent release remains incremental, facing technical capability challenges. OpenAI has, in product terms, already released a personal-use AI agent through ChatGPT agent on July 17, 2025 [15]. If the market contract implies a more autonomous, persistent, always-on personal assistant, no official OpenAI release date has been identified in retrieved sources [15]. A cautious forecast suggests continued incremental rollout during late 2026 or 2027, rather than a single announced date [15]. An official consumer personal-agent capability is most likely to be announced or begin a limited paid rollout around OpenAI DevDay on September 29, 2026, or by October 31, 2026 [1][13][9][10]. However, a broad, fully general release is less certain due to OpenAI's demonstrated pattern of staged rollouts by plan, region, and capability [1][13][9][10]. Research literature in 2026 treats ChatGPT Agent as operational, but benchmarks such as PAUSE report state-of-the-art proprietary models below 70% task completion in stateful/configuration-aware scenarios, while usability research identifies premature trust and weak recovery as ongoing problems for genuinely personal, persistent, state-aware assistance [16][17][15].
Sources (17)
  1. 1When will OpenAI release a personal AI agent?kalshi.com
  2. 2When will OpenAI release a personal AI agent? - Kalshi Odds | CoinRithmcoinrithm.com
  3. 3Introducing ChatGPT agent: bridging research and actionopenai.com
  4. 4ChatGPT agent - release notes | OpenAI Help Centerhelp.openai.com
  5. 5ChatGPT agent | OpenAI Help Centerhelp.openai.com
  6. 6AIA Forecaster: Technical Reportarxiv.org
  7. 7TimeSeek: Temporal Reliability of Agentic Forecastersarxiv.org
  8. 8Prediction Arena:Benchmarking AI Models on Real-World Prediction Marketsarxiv.org
  9. 9ChatGPT — Release Noteshelp.openai.com
  10. 10ChatGPT is now a partner for your most ambitious work | OpenAIopenai.com
  11. 11Introducing workspace agents in ChatGPTopenai.com
  12. 12The era of the personal agent has finally arrivedaxios.com
  13. 13OpenAI Plans To Introduce Managed Agents At DevDay...forbes.com
  14. 14OpenAI Could Soon Launch "o", An Always-on AI Assistantofficechai.com
  15. 15[2607.27354] PAUSE: A User-Centric Benchmark for Personal AI Assistants in Unified Service Environmentsarxiv.org
  16. 16The 2025 AI Agent Index Documenting Technical and...arxiv.org
  17. 17Industry Aspirations vs. User Realities with AI Agentsarxiv.org

2. Price Chart

Historical Price (Probability)

Outcome probability
Date

3. Significant Price Movements

Notable price changes detected in the chart, along with research into what caused each movement.

Outcome: Before Jan 1, 2027

📈 September 26, 2026: 54.0pp spike

Price increased from 24.0% to 78.0%

What happened: The primary driver of the price movement was the emergence of reports, likely fueled by social media activity and tech communities, detailing an unconfirmed OpenAI product codenamed "o," an "always-on agent" [1]. These reports cited app/configuration references, indicating a potential announcement at OpenAI's DevDay on September 29, 2026 [1]. This information, circulating on or before September 26, 2026, significantly increased market confidence in a substantial new personal AI agent release before January 1, 2027. Social media was a primary driver, disseminating these early, unofficial details and leading the market's rapid repricing.

📉 September 25, 2026: 61.0pp drop

Price decreased from 85.0% to 24.0%

What happened: The primary driver of the 61.0 percentage point drop in the prediction market on September 25, 2026, was likely a traditional news report that shifted expectations regarding OpenAI's immediate agent strategy. On September 21, 2026, Forbes reported that OpenAI planned to introduce "managed agents" at its DevDay, scheduled for September 29, 2026 [2]. This report, preceding the price movement, likely led participants to believe a specific consumer-facing "personal AI agent" release before January 1, 2027, was less probable than previously assumed, as "managed agents" might be perceived as a different offering. This traditional news announcement appeared to LEAD the price move. Social media activity was irrelevant as a primary driver for this particular market movement.

Outcome: Before Nov 1, 2026

📈 September 24, 2026: 51.0pp spike

Price increased from 3.0% to 54.0%

What happened: The primary driver of the 51.0 percentage point spike on September 24, 2026, was likely recent reporting and leaks concerning a new OpenAI product, informally named "o," an always-on personal AI assistant [1][3][4]. These reports indicated that this consumer-facing agent, which would fulfill the market's stringent definition, was expected to be announced at OpenAI's September 29, 2026 DevDay [1][3][4]. This timing provided a strong signal of release before the November 1, 2026 deadline, directly boosting market confidence [5][6][7]. While no specific social media posts from key figures were identified as the origin, such leaks and news reports would have almost certainly been widely disseminated and amplified across social media platforms, making social media a contributing accelerant rather than the primary, explicit driver.
Sources (7)
  1. 1OpenAI to announce "o" always-on agent during DevDaytestingcatalog.com
  2. 2OpenAI Plans To Introduce Managed Agents At DevDay...forbes.com
  3. 3OpenAI Could Soon Launch "o", An Always-on AI Assistantofficechai.com
  4. 4OpenAI plans new AI assistant as competition in the personal AI space intensifiesfirstpost.com
  5. 5When will OpenAI release a personal AI agent?kalshi.com
  6. 6When will OpenAI release a personal AI agent? - Kalshi Odds | CoinRithmcoinrithm.com
  7. 7When will OpenAI release a personal AI agent?kalshi.com

4. Market Data

Contract Snapshot

This Kalshi market features contracts on when OpenAI will release a personal AI agent. For each contract, a "YES" resolution is triggered if OpenAI releases a personal AI agent before the specific date stated in the contract. A "NO" resolution is triggered if the agent has not been released by that specific date.

Key deadlines for the contracts are October 1, 2026, November 1, 2026, December 1, 2026, and January 1, 2027. The page does not provide explicit definitions for what constitutes a "release" or a "personal AI agent," nor does it detail any specific settlement conditions beyond these dates.

Available Contracts

Market options and current pricing

Outcome bucket Yes (price) No (price) Last trade probability
Before Oct 1, 2026 $0.69 $0.45 69%
Before Nov 1, 2026 $0.87 $0.30 80%
Before Dec 1, 2026 $0.80 $0.26 74%
Before Jan 1, 2027 $0.78 $0.24 78%

Market Discussion

A prediction market implies an approximate 80% probability that OpenAI will release a qualifying personal AI agent, defined as a general-purpose, multi-step agent live for users outside the company, before November 1, 2026 [1][2]. Current commentary suggests a possible announcement around September 29, 2026, regarding rumored always-on agents like “o,” “Aeon,” or “Codex Bot,” building on existing agentic capabilities such as ChatGPT agent and Pulse [3][4][5][6][7][8]. However, the broader AI ecosystem and user expectations highlight significant challenges in developing agents with persistent memory and cross-application task capabilities, suggesting that adoption and trust risks may lead to staged rollouts rather than a single, clearly dated launch [9][10][11][12][13].

Sources (13)
  1. 1When will OpenAI release a personal AI agent?kalshi.com
  2. 2When will OpenAI release a personal AI agent? - Kalshi Odds | CoinRithmcoinrithm.com
  3. 3Introducing ChatGPT agent: bridging research and actionopenai.com
  4. 4Introducing ChatGPT Pulse | OpenAIopenai.com
  5. 5OpenAI's rumored 'o' agent points to a long-running Grok Bot-like productruntimewire.com
  6. 6OpenAI Aeon leak: what is actually verified before DevDayorcarouter.ai
  7. 7OpenAI Could Soon Launch "o", An Always-on AI Assistantofficechai.com
  8. 8OpenAI’s Rumored “Codex Bot” Could Be Its Answer to Grok Bottechtrendsnewsupdate.substack.com
  9. 9From AI Assistant to Personal AI: Social Intelligence as the Next Consumer Frontier - ChatGPT / Feature requests - OpenAI Developer Communitycommunity.openai.com
  10. 10The Shift to Agentic AI: Evidence from Codexarxiv.org
  11. 11CocoaBench: Evaluating unified digital agents in the wildarxiv.org
  12. 12[2602.00773] "Please, don’t kill the only model that still feels human": Understanding the #Keep4o Backlashar5iv.labs.arxiv.org
  13. 13[2609.15482] When AI Companions Disappear: Relational Continuity and Collective Contestation during China's National AI Regulatory Transitionarxiv.org

5. Trust Index

Octagon Trust Index Kalshi 75 Good

Order book is critically thin.

Primary risk· Trade quality

How it adds up
Integrity80% of score82Good
Trade quality20% of score48High Risk

Includes the cost to trade: a $10,000 order can't be filled here because the order book is too thin.

Trust score75Good

Weighted blend with hard caps — a critically weak safety pillar, or a severe trading anomaly, caps the total regardless of the rest. Full methodology · About the Trust Index

Trust profile
Integrity4 screens run · 6 don't apply
Market integrity
76Good

Outcome is hard to influence

Large tradesTrade signals · medium confidence

Largest single print on KXOAIPERSONALAGENT-27JAN01-26DEC01 carried 47% of its $2,714 tape (window is 100% of 24h flow; print is 2d old)

Screened clean
Thin-volume movesMoves carried volume Clear
Size patternsNo distribution read Clear
2 screens not run
Outcome control · doesn't apply to this market type
Thin-leg conviction · doesn't apply to this market type
Info fairness
80Good

Structural prior — no screens apply

No suspicious activity on any screen

Screened clean
Information exposureNo concentrated group can know the outcome materially early. Clear
4 screens not run
Move vs. catalyst · doesn't apply to this market type
Longshot repricing · doesn't apply to this market type
Pre-broadcast certainty · doesn't apply to this market type
Cross-venue timing · doesn't apply to this market type
Resolution quality
92Strong

Resolution mechanism is dependable

No suspicious activity on any screen

Resolution rules read: Clear

Liquidity
33Very thin

Hard to fill near the screen price

4 contracts · worst first

Before Oct 1, 20266
Very thin · high confidence

Typical bar range 27.4% of price over 4 traded bars

Before Jan 1, 202742
Thin · low confidence

Light recent trading — size may not fill near the screen price.

Before Dec 1, 202665
Tradeable · low confidence

Light recent trading — size may not fill near the screen price.

Before Nov 1, 202684
Deep · high confidence

Typical bar range 1.5% of price over 4 traded bars

24h traded notional $3,959 · 24h
Book spread 23% vs realized 1%
Move quality
92Confirmed

Moves came with real volume

4 contracts · worst first

Before Oct 1, 202691
Confirmed · medium confidence

Volume confirms the move over 24h

Before Nov 1, 202692
Confirmed · medium confidence

Volume confirms the move over 24h

Before Jan 1, 2027—
No trading in 2d

No trades in the last 2d — nothing to confirm

Before Dec 1, 2026—
No significant move (6h)

Price was stable over the last 6 hours.

Rule clarity
80Clear

Counts toward Resolution quality, not here

4 contracts · worst first

Before Nov 1, 202680
Clear · low confidence

Settles on a clear result. Verify the settlement source and resolution date.

objective outcome · Clearofficial source · Cleartime boundary · Unclear
Before Jan 1, 202780
Clear · low confidence

Settles on a clear result. Verify the settlement source and resolution date.

objective outcome · Clearofficial source · Cleartime boundary · Unclear
Before Dec 1, 202680
Clear · low confidence

Settles on a clear result. Verify the settlement source and resolution date.

objective outcome · Clearofficial source · Cleartime boundary · Unclear
Before Oct 1, 202680
Clear · low confidence

Settles on a clear result. Verify the settlement source and resolution date.

objective outcome · Clearofficial source · Cleartime boundary · Unclear
Trade quality
48High Risk

Order book is critically thin

Tradable, but expect execution cost beyond small sizes.

Outcome quality 62 (notional-weighted across 4 scored legs)

$100 order: 9.3c slippage (~12% of a 78.5c contract)

$1,000 order: 9.8c slippage (~12% of a 78.5c contract)

$10,000 order: book too thin to fill

Before Nov 1, 2026Primary87Market scoreLiquidity84Move quality92Rule clarity80
Why 87
Liquidity
84 · Deep · high confidence

Typical bar range 1.5% of price over 4 traded bars

24h traded notional $3,959 · 24h
Book spread 23% vs realized 1%
Move quality
92 · Confirmed · medium confidence

Volume confirms the move over 24h

Rule clarity
80 · Clear · low confidence

Settles on a clear result. Verify the settlement source and resolution date.

objective outcome · Clearofficial source · Cleartime boundary · Unclear
Before Jan 1, 202742Market scoreLiquidity42Move quality–Rule clarity80
Why 42
Liquidity
42 · Thin · low confidence

Light recent trading — size may not fill near the screen price.

Move quality
No trading in 2d

No trades in the last 2d — nothing to confirm

Rule clarity
80 · Clear · low confidence

Settles on a clear result. Verify the settlement source and resolution date.

objective outcome · Clearofficial source · Cleartime boundary · Unclear
Before Dec 1, 202665Market scoreLiquidity65Move quality–Rule clarity80
Why 65
Liquidity
65 · Tradeable · low confidence

Light recent trading — size may not fill near the screen price.

Move quality
No significant move (6h)

Price was stable over the last 6 hours.

Rule clarity
80 · Clear · low confidence

Settles on a clear result. Verify the settlement source and resolution date.

objective outcome · Clearofficial source · Cleartime boundary · Unclear
Before Oct 1, 202631Market scoreLiquidity6Move quality91Rule clarity80
Why 31
Liquidity
6 · Very thin · high confidence

Typical bar range 27.4% of price over 4 traded bars

Move quality
91 · Confirmed · medium confidence

Volume confirms the move over 24h

Rule clarity
80 · Clear · low confidence

Settles on a clear result. Verify the settlement source and resolution date.

objective outcome · Clearofficial source · Cleartime boundary · Unclear

underwriting_v1.4 · computed Sep 26, 2026

6. What specific announcements at OpenAI's September 29 DevDay could confirm a personal agent release before November 1, 2026?

Required Release TypeLive product in ChatGPT (not developer API) [1][2][3]
DevDay DateSeptember 29 [1][2][3]
Agents API BetaSeptember 10 [1][2][3]
OpenAI must announce a live personal agent integrated into ChatGPT. To confirm a personal AI agent release before November 1, 2026, announcements at the September 29 DevDay must clearly indicate a live product integrated directly into ChatGPT, differentiating it from a general developer API [1][2][3]. Key requirements for such confirmation include a demonstration of persistent, user-scoped memory across sessions and the agent's capability to independently execute multi-step workflows across third-party applications, complete with user-facing controls [2][4][5]. Furthermore, a consumer-facing ChatGPT interface or control, a clear rollout scope, and an actual availability date or immediate live rollout are essential [2][3].
Developer-focused releases alone will not confirm a personal agent launch. Conversely, technical documentation, a developer SDK, or a standalone API would not meet the market's expectations for a personal agent deployment [2][3]. The Agents API, a developer platform, was already in public beta as of September 10 [1][2][3]. While elements like a managed agent offering a secure execution environment, skills/plugins, and support for long-running tasks would enhance confidence in OpenAI's platform capabilities, these features are insufficient unless directly linked to a live, general-purpose personal agent accessible within ChatGPT [2][3]. An unconfirmed rumor also suggests a new 'o' agent for long-running tasks may be unveiled, though specific details on its access and capabilities remain unconfirmed [6].
Sources (6)
  1. 1OpenAI DevDay 2026devday.openai.com
  2. 2Introducing the Agents API | OpenAIopenai.com
  3. 3Agent Environment: Proven DevDay 2026 Risks to Avoidprogressiverobot.com
  4. 4Introducing apps in ChatGPT and the new Apps SDK | OpenAIopenai.com
  5. 5The next evolution of the Agents SDK | OpenAIopenai.com
  6. 6OpenAI's rumored 'o' agent points to a long-running Grok Bot-like productruntimewire.com

7. How does OpenAI's 'ChatGPT Work' product stack up against the Kalshi market's specific definition of a 'personal AI agent'?

Deployment Start DateJuly 9, 2026 [1][2][3][4]
Criteria SatisfiedMulti-step tasks and third-party application integration [1][2][3][5][4]
Unresolved CriteriaPersistent personal memory and autonomous task initiation [1][2][3][5][4]
ChatGPT Work meets some Kalshi personal AI agent criteria. OpenAI's product notably satisfies requirements concerning multi-step task handling and integration with third-party applications. OpenAI describes Work as an agent capable of managing long, multi-step processes, collecting context from connected applications and files, planning and executing tasks, and generating finished outputs such as documents or reports. Its planned deployment to external users, commencing July 9, 2026, across various subscription plans, addresses the "live and acknowledged" aspects of Kalshi's definition, confirming its operational availability and scope [1][2][3][4][5].
Key elements of Kalshi's definition remain unconfirmed for ChatGPT Work. Specifically, critical unresolved aspects include whether the product possesses persistent personal memory and can autonomously initiate multi-step tasks across third-party applications without explicit user-created schedules, triggers, or activation. Descriptions for ChatGPT Work highlight its functionality within specific project contexts, connected business environments, user-defined schedules, monitoring, and the necessity of approval or direction. These descriptions do not explicitly confirm that Work remembers individual users in the way specified by the definition or that it independently commences multi-step tasks without user prompts or pre-set conditions, leaving these facets of Kalshi's definition unconfirmed by the available information [1][2][3][5][4].
Sources (5)
  1. 1When will OpenAI release a personal AI agent?kalshi.com
  2. 2ChatGPT is now a partner for your most ambitious workopenai.com
  3. 3ChatGPT — Release Noteshelp.openai.com
  4. 4ChatGPT Work and Codex | OpenAI Help Centerhelp.openai.com
  5. 5ChatGPT Work: AI for Projects, Tasks, and Teamschatgpt.com

8. How do the rumored capabilities of OpenAI's personal agent compare to those of Google's Gemini and Apple's enhanced Siri?

OpenAI Agent speculated announcementSeptember 29, 2026 [1][2][3]
Gemini simultaneous tasksUp to ten [4][5][6][7]
Siri AI user betaLater in 2026 [8][9]
The rumored capabilities of leading AI personal agents highlight distinct strategic focuses among major technology companies. OpenAI's speculated agent is noted for its strength in persistence, autonomous follow-up work, and proactive continuity [10]. In contrast, Google's Gemini demonstrates strengths in multimodal perception, web and browser execution, deep Google-service integration, and efficient handling of parallel tasks [4][6]. Apple's enhanced Siri, meanwhile, prioritizes private, OS-level personal context retrieval and seamless cross-device application actions [8].
OpenAI's unconfirmed agent prioritizes long-term autonomous task execution and continuity across sessions. Referred to under various speculative names such as Persistent mode, Codex Bot, "o," or Aeon, this rumored agent is reported to continue work until explicitly stopped, generate follow-up tasks across sessions, leverage prior user interactions, and occasionally send proactive messages [10][1][2][11]. As of September 26, 2026, there is no public product page, documentation, or pricing available for this agent, though an announcement is speculated for OpenAI DevDay on September 29, 2026 [10][1][2][11][3]. Google’s Gemini ecosystem showcases a more developed multimodal assistant and agent direction, exemplified by Gemini and Project Astra. These can utilize video understanding, screen sharing, memory, and integrate with a wide array of Google services, including Search, Lens, Maps, Gmail, and Calendar, along with general interface control [4][6][7]. Project Mariner further extends these capabilities by enabling browser operations for tasks such as research, bookings, and shopping, with potential future updates allowing up to ten simultaneous tasks. Gemini's computer-use agents are subject to safety controls, initially operating only within the active browser tab and requiring final confirmation for sensitive actions [4][5][6][7].
Apple's enhanced Siri focuses on privacy and deep personal integration across its ecosystem. Introduced in June 2026, Siri AI emphasizes deep personal context and device actions, capable of searching messages, emails, and photos, understanding onscreen content, and performing systemwide actions across apps [8][9]. Siri AI maintains conversational history privately and syncs securely via iCloud [8][9]. While officially introduced, it is currently available for developer testing, with a user beta expected later in 2026 [8][9][12][13].
Sources (13)
  1. 1OpenAI's rumored 'o' agent points to a long-running Grok Bot-like productruntimewire.com
  2. 2OpenAI Aeon leak: what is actually verified before DevDayorcarouter.ai
  3. 3OpenAI Appears Set to Unveil Managed Agents at DevDay 2026, With Ads That Hand Off to Agents | claypierclaypier.com
  4. 4Google I/O 2025: Gemini as a universal AI assistantblog.google
  5. 5Google introduces Gemini 2.0: A new AI model for the agentic erablog.google
  6. 6Project Astra — Google DeepMinddeepmind.google
  7. 7Gemini 2.0 Project Astra features, Mariner browser agent detailed9to5google.com
  8. 8Apple introduces Siri AI, a profoundly more capable and...apple.com
  9. 9Apple unveils next generation of Apple Intelligence, Siri AI...apple.com
  10. 10OpenAI Is Developing a ‘Persistent’ AI Agent | WIREDwired.com
  11. 11OpenAI’s Rumored “Codex Bot” Could Be Its Answer to Grok Bottechtrendsnewsupdate.substack.com
  12. 12Apple says some AI improvements to Siri delayed to 2026 | Reutersreuters.com
  13. 13Apple delays Siri AI upgrades again as technical challenges mount | TechSpottechspot.com

9. Beyond OpenAI's official blog, which developer forums or code repositories offer the earliest signals of a public-facing agent rollout?

Agents API Public BetaSeptember 10, 2026 [1]
Codex Rollout ThreadJune 3, 2025 [2]
openai/openai-agents-python v0.8.0 ReleaseFebruary 5, 2026 [3][4][5]
OpenAI's agent rollout primarily targets developers via forum announcements. The earliest indications of a public-facing agent rollout from OpenAI are evident in its Developer Community's Announcements forum and specific GitHub repositories, pointing specifically to a developer-platform agent rather than a consumer-facing personal AI agent [1][6]. On September 10, 2026, the OpenAI Developer Community announced the public beta of the Agents API and OpenAI-hosted sandboxes for all developers [1]. This announcement detailed key features such as the Codex agent harness, long-running sessions, context management, file/code execution, and sandbox integrations [1][2][7]. A significant preceding event was the Developer Community Codex rollout thread on June 3, 2025, which indicated Codex's expansion to ChatGPT Plus, opt-in internet access, and support for GitHub-connected task execution [2].
GitHub repositories provide granular insights into agent development and features. Code-level signals are clearly visible in the openai/openai-agents-python GitHub repository, whose documentation outlines concepts like agents, handoffs, orchestration, hosted tools, and computer/shell execution, including an experimental Codex tool [3][4][5]. The v0.8.0 release of this repository, dated February 5, 2026, notably introduced human-in-the-loop approval and resumable runs [3][4][5]. Additionally, the openai/openai-realtime-agents GitHub repository presents an earlier public pattern for multimodal personal-agent behavior, encompassing reusable voice-agent configurations, tool calls, and specialist handoffs [8].
User requests for personal AI agents differ from official rollouts. While discussions about personal AI runtimes are present within the Developer Community, these represent user-proposed feature requests rather than official OpenAI rollouts [9][10][11][12][13].
Sources (13)
  1. 1Introducing the Agents API and hosted sandboxes - Announcements - OpenAI Developer Communitycommunity.openai.com
  2. 2Codex Updates: Internet access and rollout to Plus users - Codex - OpenAI Developer Communitycommunity.openai.com
  3. 3docs/agents.mdgithub.com
  4. 4docs/tools.mdgithub.com
  5. 5v0.8.0github.com
  6. 6Introducing the Agents APIopenai.com
  7. 7The next evolution of the Agents SDK | OpenAIopenai.com
  8. 8README.mdgithub.com
  9. 9Feature Concept: A ChatGPT Personal AI Runtime — Wearables, Local Compute, and an Open Device Layer - ChatGPT / Feature requests - OpenAI Developer Communitycommunity.openai.com
  10. 10AI Personal Agent with granular app permissions - ChatGPT / Feature requests - OpenAI Developer Communitycommunity.openai.com
  11. 11An Ambient, Proactive Personal Assistant Layer for ChatGPT - ChatGPT / Feature requests - OpenAI Developer Communitycommunity.openai.com
  12. 12Make ChatGPT My Single, Persistent Personal AI Across Devices and Services - ChatGPT / Feature requests - OpenAI Developer Communitycommunity.openai.com
  13. 13Feature Proposal: ChatGPT as the Personal AI Infrastructure Layer — Not Another Device - ChatGPT / Feature requests - OpenAI Developer Communitycommunity.openai.com

10. What does the rollout history of OpenAI's enterprise and workspace agents suggest about the timeline for a consumer-facing personal agent in Q4 2026?

Plausible Launch WindowQ4 2026 [1][2][3]
Most Likely Release WindowOctober-November 2026 [4][5][3]
Agents API Public BetaSeptember 10, 2026 [4]
OpenAI's rollout history suggests a plausible Q4 2026 consumer agent launch. The company's agent capabilities indicate that a consumer-facing personal agent is plausible in Q4 2026, particularly as an initial paid-tier or limited rollout, rather than a fully mature, broadly available, always-on personal agent [1][2][3]. The most likely window for such a release is October-November 2026, with an initial US paid-user rollout or waitlist being more probable than universal availability [4][5][3]. This timeline is supported by OpenAI's pattern of staged but rapid agent rollouts, as exemplified by the ChatGPT agent, which launched to Pro users on July 17, 2025, and expanded within days or weeks [6][3][7].
Enterprise rollouts and API availability present a contrasting, more cautious timeline. Workspace agents, for example, followed a more conservative enterprise path, undergoing gradual rollout over several weeks to achieve general availability [1][8][2]. The Agents API, which exposes the underlying agent infrastructure, was introduced in public beta on September 10, 2026, suggesting that the platform infrastructure is ahead of fully mature consumer autonomy [4]. While consumer building blocks such as ChatGPT memory and ChatGPT Work's ability to act across applications are in place, these are separate capabilities, not an explicitly announced always-on personal agent [5][9][3]. A December 2026 launch remains possible if features like memory, autonomous triggers, third-party permissions, safety, and reliability are considered blockers for general availability [4][5][3]. No official source currently announces a specific date for a personal agent, and the Kalshi contract narrowly defines a qualifying event as a general-purpose personal agent that remembers the user and starts multi-step tasks across third-party apps, explicitly excluding beta or preview access [10].
Sources (10)
  1. 1Introducing workspace agents in ChatGPT | OpenAIopenai.com
  2. 2ChatGPT Business - Release Notes | OpenAI Help Centerhelp.openai.com
  3. 3ChatGPT is now a partner for your most ambitious work | OpenAIopenai.com
  4. 4Introducing the Agents API | OpenAIopenai.com
  5. 5Memory FAQ | OpenAI Help Centerhelp.openai.com
  6. 6Introducing ChatGPT agent: bridging research and actionopenai.com
  7. 7ChatGPT agent | OpenAI Help Centerhelp.openai.com
  8. 8ChatGPT Enterprise and Edu release noteshelp.openai.com
  9. 9Memory FAQ | OpenAI Help Centerhelp.openai.com
  10. 10When will OpenAI release a personal AI agent?kalshi.com

11. What Could Change the Odds

Key Catalysts

OpenAI launched the ChatGPT agent on July 17, 2025, initially for Pro, Plus, and Team users, with subsequent rollout to Enterprise and Education accounts [1][2][3]. This agent integrates Operator-style web interaction with deep research, enabling multi-step tasks such as web research, form filling, spreadsheet editing, travel planning, and appointment scheduling [1]. This established foundation, its personal-use cases, and broad paid-plan deployment represent bullish catalysts for a November 1, 2026-or-earlier resolution [4][1][3]. Additionally, the Agents API entered public beta on September 10, 2026, indicating continued expansion of agent infrastructure for developers [5].
The Kalshi market definition, however, is narrower, requiring a live, officially acknowledged OpenAI product in the United States that remembers the user and initiates multi-step tasks across third-party applications, specifically excluding beta, test, experimental, preview, or early-access releases [4]. This strict exclusion, alongside potential safety and reliability constraints, staged rollouts, and usage limits, acts as a bearish catalyst [4][1][3]. OpenAI’s product strategy emphasizes incremental convergence and iterative capability improvements rather than a single future launch [1][6]. While the Agents API expands developer capabilities, it is not evidence of a new consumer personal-agent launch that would satisfy the Kalshi criterion [5]. As of September 26, 2026, no retrieved source announced a future dated standalone consumer “personal AI agent” release [7][8][9][10].

Key Dates & Catalysts

  • Expiration: October 08, 2026
  • Closes: January 01, 2027
Sources (10)
  1. 1Introducing ChatGPT agent: bridging research and actionopenai.com
  2. 2Introducing ChatGPT agent: bridging research and action | OpenAIopenai.com
  3. 3ChatGPT agent - release notes | OpenAI Help Centerhelp.openai.com
  4. 4When will OpenAI release a personal AI agent? - Kalshi Odds | CoinRithmcoinrithm.com
  5. 5Introducing the Agents API | OpenAIopenai.com
  6. 6Introducing deep researchopenai.com
  7. 7What Could the Agent See at 19:05? Generating Temporal Enterprise Scenarios from Real Research and Replaying Them to Evaluate Agentsarxiv.org
  8. 8MyPCBench: A Benchmark for Personally Intelligent Computer-Use Agentsarxiv.org
  9. 9[2604.11548] SemaClaw: A Step Towards General-Purpose Personal AI Agents through Harness Engineeringarxiv.org
  10. 10PAUSE: A User-Centric Benchmark for Personal AI Assistants in Unified Service Environmentsarxiv.org

13. Related News

+54.0pp
Last updated: September 26, 2026, 13:03 UTC

OpenAI 'o' Agent Leaks Fuel Bets on Imminent Personal AI Release

Speculation around a new, unannounced OpenAI personal AI assistant, reportedly named "o," fueled a significant repricing in prediction markets on Saturday, September 26, 2026. Following online leaks a...