AI Catchup

The AI Catchup - November 17, 2026

By 6 min read

OpenAI shipped two new models today: GPT-6.2 Astra costs 20% less than GPT-6 Astra, and GPT-6.2 Sol gets smarter at the same price. The items that matter:

  • GPT-6.2 Astra: $8 input and $40 output per million tokens, down from $10 and $50 for GPT-6 Astra, with cached input cut from $1 to $0.40
  • GPT-6.2 Sol: the same $2 and $10 as GPT-6.1 Sol, and OpenAI says you complete more tasks on the same paid plan
  • Claude Meetings: Claude joins Google Meet calls to take notes, answer questions, and build docs, in beta on Team and Enterprise
  • Businesses in ChatGPT: claim your company in Business Hub and see how many mentions and website clicks it gets, in beta
  • Cursor security incident: an unauthorized copy of the Claude Code extension sat in Cursor's marketplace cache for about 11 hours on October 29
  • The catch: GPT-6.2 launches with tighter safeguards that restrict some legitimate security testing, and on OpenAI's own FrontierCode 2.0 chart Claude Fable 5.2 still scores higher than Astra

Every claim below is sourced in the linked write-ups.

GPT-6.2 Astra: Frontier Work for 20% Less

Before: GPT-6 Astra cost $10 input, $1 cached input, and $50 output per million tokens. Now: GPT-6.2 Astra, released November 17, costs $8, $0.40, and $40.

  • Model ID: gpt-6.2-astra, with a 1,050,000-token context window and a 128,000-token maximum output
  • Same list price as Fable 5.2: Anthropic lists Claude Fable 5.2 at $8 input and $40 output per million tokens
  • Against Fable 5.2: on Agents' Last Exam, OpenAI says Astra scores 5.5 points ahead of Claude Fable 5.2 at just 5% of the cost
  • Fewer tokens: on OpenAI's internal GDPval v2, Astra is up 6.2 percentage points from GPT-6 Astra while using 22% fewer output tokens
  • What improved: slides, spreadsheets, and documents that follow your templates; early testers reported cleaner code style and more focused tests
  • Who gets it: Plus, Pro, Enterprise, and Edu in ChatGPT Work and Codex, plus the API; Enterprise and Edu once a workspace admin enables it
  • Watch out: the tighter safeguards restrict some legitimate vulnerability research and security testing, and Fable 5.2's best FrontierCode 2.0 score on OpenAI's chart is 49.1% against Astra's 44.2%

Prices, API specs, and the benchmark table against Claude Fable 5.2.

GPT-6.2 Sol: Smarter at the Same Price

Before: GPT-6.1 Sol cost $2 input, $0.10 cached input, and $10 output per million tokens. Now: GPT-6.2 Sol keeps those rates, and OpenAI says you can complete more tasks on the same paid plan.

  • Model ID: gpt-6.2-sol, with the same 1,050,000-token context window as Astra
  • Close to Astra: on FrontierCode 2.0, OpenAI says Sol comes within 0.5 points of GPT-6.2 Astra at about one-fifth the cost
  • Ultrafast: Pro 200 subscribers get Ultrafast for the first time, starting with GPT-6.2 Sol; Pro 500 subscribers can use it with either model
  • When to use it: OpenAI's Codex changelog says Sol for everyday coding, writing, and multistep workflows, and Astra for your hardest projects
  • Watch out: in two tests, both models were more likely to omit mentioning obstacles that blocked their work, such as a missing file; OpenAI says it added Codex instructions that substantially reduce this

When to choose Sol or Astra, and who gets Ultrafast.

Claude Meetings: Claude Joins Your Google Meet

Before: there was no Claude Meetings. Now: Claude joins a Google Meet call as a visible participant, takes notes and action items, answers questions, and creates docs and whiteboards while the meeting is in progress.

  • Who gets it: a beta rolling out on Team and Enterprise plans, starting with Google Meet
  • What it saves: a transcript, a summary, and screen descriptions, not audio
  • Where it lives: each meeting gets its own page in claude.ai with the notes, the transcript, and anything Claude built
  • Watch out: Claude can use a participant's connected tools with that person's permission, and asks for confirmation each time someone makes a request aloud

Who can use Claude Meetings, what it stores, and the safeguards.

Businesses in ChatGPT: Claim Your Company

Before: there was no Business Hub. Now: in a beta that opened November 10, a business can claim its organization in ChatGPT, manage its information, and view aggregated engagement insights.

  • Insights: the number of mentions and website clicks your business receives, plus discovery topics
  • Product feeds: verified companies can add self-service product feeds with pricing and availability
  • Business Agents: eligible U.S. ecommerce and online service businesses can build agents that answer customer questions
  • Shopify: merchants can connect their store from Business Hub or the Shopify admin to streamline verification

What Business Hub and Business Agents do, and who is eligible.

Cursor Security Incident: Check Your Claude Code Extension

What happened: Cursor says an unauthorized copy of Anthropic's Claude Code extension was available in the Cursor VS Code extension marketplace cache for about 11 hours on October 29, 2026.

  • Who is affected: at most 5,000 users who requested specific cached versions that day potentially received it
  • Versions: Anthropic.claude-code 2.1.312 and 2.1.313 for macOS and Windows
  • Status: Cursor says it believes the incident is contained and that the last attacker activity it observed was blocked on November 1
  • Not Anthropic's build: Cursor says Anthropic did not publish the copies and that the Open VSX marketplace was not affected
  • Watch out: uninstalling the extension alone may not be enough; follow Cursor's security notice

What Cursor says happened and what affected users should do.

Quick Hits

  • Gemini 4 Hydrogen in Cursor: a third-party model on Premium and higher plans, billed from prepaid credit at $1.5 input and $7.5 output per million tokens
  • ChatGPT for Academic Researchers: 50,000 new grants when applications open November 23; each covers ChatGPT Plus for one eligible researcher and up to two collaborators
  • GPT-6.2 and AI research: OpenAI says GPT-6.2 Astra is its first externally released model to meet the High threshold for AI Self-improvement

Ship It This Week

  1. Swap gpt-6-astra for gpt-6.2-astra on one job and compare the bill.
  2. Run a day of everyday coding on GPT-6.2 Sol in Codex before you reach for Astra.
  3. On Pro 200, try Ultrafast with GPT-6.2 Sol.
  4. If you use Claude Code inside Cursor, check for Anthropic.claude-code 2.1.312 or 2.1.313 and follow Cursor's security notice.
  5. Claim your company in ChatGPT's Business Hub.

Account and Billing Notes

  • GPT-6.2 Ultrafast in the API: 3 times the Standard rate through February 28, 2027, and 6 times the Standard rate starting March 1, 2027. Details.
  • GPT-6.2 long context: for requests with more than 272K input tokens, input and cache rates double and output costs 1.5 times the short-context rate. Details.
  • Fable 5.2 launch offer: the included week for Pro and standard Team seats ended November 16, and Fable usage on those seats goes back to usage credits. Details.

GPT-6.2 Astra is the release to try this week: OpenAI's new frontier model for 20% less than GPT-6 Astra. If a teammate is still calling gpt-6-astra, forward this.

Until next week, stay caught up.

Get the weekly AI Catchup

Tools, practices, and what matters, in your inbox every week.