This week’s AI news roundup is led by GPT-6.2 Astra, the model OpenAI launched on 17 November 2026 for demanding professional work, alongside GPT-6.2 Sol for everyday tasks at an unchanged price. A day later Google answered with Gemini 4 Neon, having already made Tasks in Gemini free in six countries.
Below are 8 official updates published between Friday 13 November and Thursday 19 November 2026. For each one you’ll find what happened, what it means for your business, and whether you need to act now.
This week in a minute
| Company | The news in one line | Announced | Who it matters to |
|---|---|---|---|
| OpenAI | GPT-6.2 Astra for hard projects and GPT-6.2 Sol for daily work; Astra’s API price drops 20% | 17 November 2026 | Businesses, developers, paid subscribers |
| Gemini 4 Neon sits between Carbon and Hydrogen at $3 per million input tokens | 18 November 2026 | Developers, analysts, Pro and Ultra subscribers | |
| Tasks in Gemini becomes free in six countries, with calling and buying features starting in the US | 17 November 2026 | Individuals and small business owners | |
| OpenAI | ChatGPT for Sales: an OpenAI Sales plugin plus 30+ partner plugins | 17 November 2026 | Sales teams and account managers |
| Gemini Notebook builds an interactive site from your sources on request | 16 November 2026 | Marketers, students, project managers | |
| Google DeepMind | Genie 4 generates interactive worlds with real-time audio for Ultra subscribers | 18 November 2026 | Content creators and designers |
| Google Search Central | Branded, multi-country and multi-device filters arrive in the Search Analytics API | 16 November 2026 | SEO teams and reporting tools |
| Meta | Direct, a standalone Instagram messaging app, enters testing in Türkiye and Thailand | 16 November 2026 | Social media and brand managers |
What’s new in GPT-6.2 Astra and Sol?
GPT-6.2 Astra is OpenAI’s new frontier model for its hardest projects, and GPT-6.2 Sol is a lighter model for everyday work. Both were announced on 17 November 2026, and they top this AI news roundup because they raise quality and cut cost at the same time.
What happened
According to OpenAI’s official GPT-6.2 Astra and Sol announcement, both models are available from launch day to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and to developers through the API (the interface apps use to call the model directly). The announcement doesn’t mention the free plan.
GPT-6.2 Astra costs $8 per million input tokens, $40 per million output tokens and $0.40 for cached input. A token is the small unit of text a model reads and writes. That’s 20% below GPT-6 Astra, with a deeper 60% discount on cached input. GPT-6.2 Sol keeps GPT-6.1 Sol’s rates of $2 input and $10 output (prices as listed on 17 November 2026; they can change).
OpenAI says Astra is noticeably better at slides, spreadsheets and documents. Give it a screenshot of one slide and it can match the layout, colours and exact font size across the rest of the deck. On OpenAI’s internal GDPval v2 benchmark it gained 6.2 points over GPT-6 Astra while using 22% fewer output tokens, and Sol came within half a point of Astra on coding at about a fifth of the cost.
What GPT-6.2 Astra means for you
The practical story isn’t just a smarter model. It’s a model that needs fewer tokens to get to the same result, so you finish more work inside your plan’s limits. Marketers who build client decks will feel it first, because GPT-6.2 Astra sticks to the company template instead of reinventing the design every time.
One caveat: OpenAI shipped both models with tighter safeguards on high-risk cybersecurity requests, and it admits these limits can block some legitimate security testing.
Do you need to do anything?
The first step this AI news roundup suggests: if you’re on a paid plan, try GPT-6.2 Sol for routine tasks first and keep GPT-6.2 Astra for long reports and complex presentations. If your team uses the API, rerun cost estimates for your largest jobs at the new prices. OpenAI’s services are generally available across the Gulf; from Syria, check OpenAI’s supported-countries list before subscribing.
What is Gemini 4 Neon, and where does it fit?
Gemini 4 Neon is a new Google model positioned between the more powerful Gemini 4 Carbon and the faster Gemini 4 Hydrogen. Launched on 18 November 2026, it’s Google’s pitch for frontier-level intelligence at a lower cost per task.
Google’s official Gemini 4 Neon announcement lists pricing at $3 per million input tokens, $15 per million output tokens and $0.15 for cached input, with a 1 million-token context window. The context window is how much text the model can take in a single request, and a million tokens is enough for long files, spreadsheets and decks together. Google says Neon’s cost per task at its default thinking level is 32% lower than Carbon’s.
Where can you use it? In the paid Gemini API through Google AI Studio, in Gemini for business, in the Gemini app for Google AI Pro and Ultra subscribers only, and in the Google Antigravity developer platform. It offers four thinking levels so you can match effort to the task.
What this AI news roundup item means for you: Pro subscribers now have a quick middle option for analysis, report writing and reading PDFs and spreadsheets. Google says Gemini 4 Neon recorded its lowest fabrication rate yet in internal testing, which matters if you rely on a model for numbers and reports. The practical step: on Pro, pick Neon from the model menu and run one real monthly report through it, then compare with what you use today.
Tasks in Gemini goes free in six countries: what changed?
Since 17 November 2026, Tasks in Gemini has been free in the United States, Japan, Korea, Canada, Australia and India, powered by Gemini 4 Hydrogen. It turns Gemini from an assistant that answers into an AI agent, software that carries out a job step by step rather than just explaining how.
In Google’s announcement of free Tasks in Gemini, Josh Woodward, Vice President of Google Labs and the Gemini app, describes Gemini calling two dealerships at once to book a service slot, analysing utility bills from email, and buying a product through Google Pay within a spending cap you set. Advanced features such as calling businesses, texting Gemini and forwarding emails to gemini@google.com start in the US.
Using it is simple: flip the home-screen toggle from Chat to Task and describe what you need. Google says each task runs in a hardware-isolated virtual machine dedicated to your account, and Gemini asks for confirmation before it presses “Buy”.
For a business owner, this part of the AI news roundup points somewhere clear: assistants will call your shop, book and buy on a customer’s behalf. The Gulf and Syria aren’t among the six countries yet, so nothing is required today. Still, make sure your phone number, opening hours and prices are correct on your Google Maps profile, because that is exactly what an agent reads when it looks for a nearby business.
ChatGPT for Sales brings plugins to deal teams
On 17 November 2026 OpenAI launched ChatGPT for Sales: an OpenAI-built Sales plugin plus more than 30 new or enhanced partner plugins. A plugin connects ChatGPT to another tool you use so it can read from it and write back.
According to OpenAI’s official ChatGPT for Sales announcement, teams can connect CRM systems such as Salesforce and HubSpot, the tools that store customer and deal records, alongside Gong for call notes, Canva and Gamma for presentations, and Docusign for agreements. Admins on Business and Enterprise plans switch it on from the Admin Console under Plugins. OpenAI notes that partner plugin availability varies by plan and region.
Of everything in this AI news roundup, this is the most hands-on update for sales teams. A seller can ask for a pre-meeting brief that pulls past calls and deal status together, then draft the follow-up and update the customer record in the same conversation. If your team runs HubSpot or Salesforce with ChatGPT Business, pilot it on one deal before rolling it out, and review the permissions each plugin gets.
Gemini Notebook turns your sources into an interactive site
Since 16 November 2026, Gemini Notebook (the tool known as NotebookLM until July) can turn the sources in a notebook into an interactive site. Type “make me a site” in the chat with a short description of what you want.
Google’s post on interactive sites in Gemini Notebook gives examples such as an interactive periodic table, a dashboard for quarterly sales and a project tracker with deadlines. The feature is rolling out globally on the web to users aged 18 and over, including Google Workspace subscribers, and is coming to the mobile app.
For marketers, it’s one of the quickest wins in this AI news roundup: a fast way to turn a monthly report or market research into a page a client can explore instead of a long PDF. Try it on one report and check every number on the resulting page before you share it, since it’s built from your sources and has to match them.

Genie 4 opens to Google AI Ultra subscribers
On 18 November 2026 Google began rolling out a new Project Genie experience to eligible Google AI Ultra subscribers, giving first access to Genie 4. Genie is a world model: it generates a 3D environment you can move through and interact with, not a fixed image or video.
According to Google’s announcement that Genie 4 is now available, you can type a new prompt mid-session to change the weather or drop in a character, with native audio that adapts to the scene. Genie 3 now runs up to 3 minutes per generation instead of 60 seconds and can start from a 360-degree panorama. There’s a public gallery for sharing and remixing worlds, and all video and audio carry SynthID, an invisible watermark that marks content as AI-generated.
In this AI news roundup, Genie 4 is for experimenting rather than daily work, and Google AI Pro members can only join the waitlist. Still, content creators and ad teams may find it handy for quick prototypes of an interactive experience or a virtual tour before committing to a full production.
SEO and social updates: Search Console and Instagram’s Direct
Roughly a quarter of this AI news roundup comes from search and social platforms, with two official updates worth tracking. Paid advertising was quiet: we found no major official Google Ads or Meta ads announcement we could verify on their own blogs during the period.
The Search Analytics API gets new filters
On 16 November 2026 Google announced on the Google Search Central Blog that three capabilities from the Search Console Performance report are now in the Search Analytics API: the branded queries filter, web multimodal search reporting, and multi-country and multi-device filtering in a single request.
In plain terms, you can now pull figures for a group of countries, such as the whole GCC, in one call, and automatically separate searches for your brand name from generic searches. That makes automated reports more accurate and shows whether growth comes from your reputation or your content. If you use Looker Studio or another API-based report, ask whoever maintains it to switch to the new filters.
Meta tests Direct, a standalone Instagram messaging app
Meta announced in the Meta Newsroom that Direct, a messaging-only app from Instagram, is being tested in Türkiye and Thailand on Android and iOS. It works with or without an Instagram account (sign-up by email or phone), chats sync with Instagram, and the minimum age is 13.
The post says nothing about business accounts or ads. So nothing changes in how you handle customer messages today, but it’s worth watching: if Direct expands, part of customer conversation on Instagram could move into a separate app.
Where are we now? Reading this AI news roundup
Everything above comes from the companies’ own announcements. What follows is our analysis and forecasts, and some of it may prove wrong.
First, the competition has shifted from “who has the smartest model” to “who delivers the same intelligence at the lowest cost per task”. OpenAI cut Astra’s price and held Sol’s; Google pitched Neon as cheaper per task than Carbon. The upshot is that tools once reserved for large companies are moving within reach of small-business budgets.
Second, assistants now act instead of just answering. Tasks in Gemini calls and buys, and the ChatGPT Sales plugin updates customer records. At DMANTRA we notice that accurate, consistent business information (phone, address, opening hours, prices) is exactly what these agents read when they choose a shop or service for a customer. We expect keeping that data current to become a routine marketing task through 2027.
Third, on search: Google’s November core update still hadn’t been marked complete on the Search Status Dashboard as of the morning of 20 November. For context on what came before, see last week’s roundup on Gemini 4 Hydrogen and the core update, and for the bigger picture on rankings and AI answers, read our guide to modern SEO in the age of AI answers.
What to do after this AI news roundup
If you only act on two items from this AI news roundup, make them these: update your business details on Google Maps and your website, because AI agents read them, and test GPT-6.2 Sol or Gemini 4 Neon on one recurring task in your team to see how much time it really saves.
DMANTRA, a Damascus-based digital marketing agency serving clients in Syria, the Gulf and worldwide remotely, tracks these updates every week and turns them into practical decisions on SEO, content and campaigns. If you want your site ready to show up in search and in AI answers, see our SEO and GEO service, read how we work on digital marketing for Gulf businesses, or contact our team and we’ll start from your own data.
Frequently asked questions about this AI news roundup
What is the difference between GPT-6.2 Astra and GPT-6.2 Sol?
GPT-6.2 Astra is the frontier model for complex projects such as long reports, presentations and hard coding, at $8 input and $40 output per million tokens. GPT-6.2 Sol is a lighter everyday model at $2 and $10, and OpenAI says it comes close to Astra on coding at about a fifth of the cost.
Is GPT-6.2 Astra free?
OpenAI’s announcement doesn’t mention the free plan. Both models are available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and to developers through the API.
Who can use Gemini 4 Neon in the Gemini app?
Only Google AI Pro and Ultra subscribers, from the model menu. Developers can use it on the paid tier of the Gemini API through Google AI Studio, and businesses through Gemini for business.
Is free Tasks in Gemini available in the UAE, Saudi Arabia or Syria?
Not yet. Google made it free in the United States, Japan, Korea, Canada, Australia and India, and advanced features such as calling and buying start in the US.
Does Instagram’s Direct app affect my business account messages?
Meta’s announcement mentions no change for business accounts, and Direct is only being tested in Türkiye and Thailand. Customer messages on Instagram continue to work as before.







