Model intelligence.
Follow the changes. Understand what comes next.
OpenAI Launches GPT-6.2 Astra and GPT-6.2 Sol Across ChatGPT Work, Codex, and the API
At 01:45:00 Asia/Shanghai on November 18, 2026 (OpenAI News RSS; confirmed on @OpenAI at 01:54:03 Asia/Shanghai), OpenAI launched flagship GPT-6.2 Astra (gpt-6.2-astra) and everyday GPT-6.2 Sol (gpt-6.2-sol) on ChatGPT Work, Codex, and the API, bringing Ultrafast to Pro 200 starting with Sol, plus the gpt-6.2-astra-law configuration.
Latest stories116
Newest publishedExplore this company’s models & APIs ↗Explore provider coverage ↗
OpenAI Expands Online Child Safety Tools with Free Vetted API Classifier and ChatGPT Interventions
On November 16, 2026, building on its Child Safety Blueprint, OpenAI announced four online child safety updates: free OpenAI API access for vetted organizations to a new image classifier with an AI-generated signal, in-product ChatGPT warnings it is building, plus a forthcoming 24-hour access pause, new developer guidance, and investigative tools and training for authorized law enforcement.
OpenAI Expands ChatGPT for Academic Researchers to Up to 200,000 Scientists and Releases Science Usage Report
On November 16, 2026, OpenAI announced an expansion of its ChatGPT for Academic Researchers program, doubling its overall target to up to 200,000 researchers worldwide (also described in the accompanying report as more than 200,000). Applications for 50,000 new first-come, first-served grants open on Monday, November 23, 2026, providing free ChatGPT Plus access for each recipient and up to two collaborators (supporting up to 150,000 researchers) alongside the new Science at the Frontier report.
OpenAI Launches First Phase of Embedded Safety Assessments with Apollo Research, METR, Transluce, and Goodfire
On November 13, 2026, OpenAI announced agreements with four independent assessors—Apollo Research, METR, Transluce, and Goodfire—for its first phase of embedded assessments. Over extended periods, external experts receive scoped access to internal checkpoints, code, training/evaluation data, and monitoring systems to evaluate model alignment and safeguards. Here is how each assessor's scope, oversight, and conflict-of-interest rules are structured.
OpenAI Introduces Computer-Use Agents, Upgraded Clinical Search, and 20+ Healthcare Plugins for ChatGPT
On November 16, 2026, OpenAI announced new capabilities across ChatGPT for Healthcare and ChatGPT for Clinicians (free for verified U.S. clinicians), introducing computer-use agents that navigate existing healthcare software, improved trusted clinical search, and more than 20 new healthcare plugins. OpenAI also published workflow evaluations for GPT-6 Astra and GPT-6 Luna alongside HIPAA/BAA governance and human-in-the-loop review controls.
OpenAI Launches Dedicated Science Blog Alongside Explainer on Candidate Lattice Definition of the Standard Model
On November 16, 2026 (Beijing time), OpenAI launched a new blog for science and published an explainer detailing physicist Nathaniel Craig's October 6, 2026 preprint (arXiv:2610.09093), developed with GPT-6 Astra, proposing a candidate nonperturbative lattice definition of the Standard Model. This is a scientific communication release, and key questions around reflection positivity and continuum spacetime remain open.
Troubleshooting OpenAI Responses API: Using input_transformations to Catch Dropped Reasoning Items
Released on November 12, 2026, OpenAI Python SDK v3.50.0 (alongside Node SDK v7.56.0) adds the input_transformations array to Responses API response objects. This field exposes which encrypted reasoning items were withheld before model execution and why—covering prefix_mismatch, turn_boundary, model_mismatch, organization_mismatch, and end_user_mismatch. This guide explains how silent reasoning drops happen during multi-turn history edits, how to trace prefix_mismatch using item IDs, and why Prompt Cache hits do not guarantee reasoning reuse.
OpenAI Introduces Businesses in ChatGPT in Beta
OpenAI announced Businesses in ChatGPT in beta in the early hours of 11 November 2026 (Beijing time). Eligible businesses can claim profiles, update information, and view aggregated insights in Business Hub. Business Agents are currently available to eligible ecommerce and online service businesses in the United States. User conversations are not shared with businesses, and publishing an Agent does not authorize paid advertising.
Codex CLI 0.167.1 Stable Version Is Published, and npm Latest Points to It
OpenAI released Codex CLI stable version 0.167.1 on November 11, 2026 (Beijing time), tagged rust-v0.167.1 on GitHub. npm @openai/codex latest points to this version. The release notes link to the comparison with 0.167.0 without listing individual feature changes. The separate npm alpha channel is at 0.169.0-alpha.9.
Odessia's Founder Announces Acquisition by OpenAI, Will Join to Focus on the Future of Travel
Odessia's site says “OpenAI has acquired Odessia” and the service is no longer available; bookings: support@odessia.com. Founder Francis Davidson (@FDavidsonT) wrote in “OpenAI Acquires Odessia” on X (2026-11-10 18:00:11 UTC, 2026-11-11 02:00:11 Asia/Shanghai) that he will join the team to focus on the future of travel. LinkedIn: same opening, Published Nov 10, 2026 only. Greg Brockman (@gdb) replied (2026-11-10 20:48:04 UTC, 2026-11-11 04:48:04 Asia/Shanghai): “welcome to the team!” OpenAI's newsroom, @OpenAI and @OpenAIDevs have not posted. Terms, price, and ChatGPT plans are undisclosed.
OpenAI Introduces Control Room to Track and Halt Research Model Workloads
According to OpenAI (@OpenAI) on X at 2026-11-10 06:01:00 Asia/Shanghai (UTC 2026-11-09T22:01:00.000Z), Safety blog Nov 9, 2026: Control Room tracks research-infra model workloads and can stop them quickly on safety/security concerns. Misalignment monitoring may send a stop (job_id). After a stop, output stops within ~35s (auth-cache lag; goal 0). Already on research infrastructure; external serving has separate takedown. Fail-closed denies inference if Control Room is down—live for all frontier research; other research rolling. Not a ChatGPT kill switch. We haven't tested Control Room.
Same ChatGPT Dot Can Connect Multiple Local Computers and Choose by Task
According to OpenAI Developers (@OpenAIDevs) on X at 2026-11-10 04:01:49 Asia/Shanghai (UTC 2026-11-09T20:01:49.000Z), the same ChatGPT Dot can connect multiple local computers and choose which runs a task—build/test on a dedicated machine while coding continues on a laptop. Examples: second-Mac Xcode + screenshots; laptop vs older-SDK tests; second-computer UI breakpoints. Docs: Allow → Allow access; repeat; one computer/thread; cloud stays; phone reaches authorized local; local Work/Codex. @OpenAI RT. Not Dot launch, not iOS UX, not Codex delegate. We haven't tested multi-computer connect.
ChatGPT Enterprise Now Available in AWS Marketplace
According to OpenAI’s official website news (November 9, 2026), ChatGPT Enterprise is on AWS Marketplace (prodview-sscra4m7iracg). Buy via existing AWS billing; eligible spend may count toward EDP/PPA (per agreement). Package: Chat, ChatGPT Work, Codex, dots—same models/features as direct from OpenAI. OpenAI says products complement Bedrock OpenAI models, not replace them. Eligible customers may get Healthcare/FS via Marketplace. Listing: data not used to train by default; Marketplace purchase does not guarantee all data stays on AWS.