Saazora
SaaS News
Nov 16, 2026 14 mins read 521 views

AI & SaaS News, Nov 16: Gemini Wallet, Cursor Breach

Gemini now reads Google Wallet, Cursor closes its breach probe, AI API deadlines hit Nov 30 to Dec 11, and Anthropic prices its IPO this week.

Quan Le
Quan Le
AI Tools Reviewer
AI & SaaS News, Nov 16: Gemini Wallet, Cursor Breach

Last checked November 16, 2026 · News roundup

Today in brief: Gemini can now read your Google Wallet and, if you link them through Plaid, your bank and card accounts, while Cursor finished its breach investigation and told affected users which keys to rotate. Several AI API prices and shutdowns land between November 30 and December 11, so check your model names this week. On the business side, Anthropic prices its IPO Wednesday, Nvidia reports Tuesday amid a petition to bring back Jensen Huang, and Cloudflare added hard spending caps.

AI

Gemini + Google Wallet · API deadlines Nov 30–Dec 11 · Cursor breach update · Grok 4.8 rollout · OpenAI robotics and Odessia

SaaS & business

Cloudflare spending caps · Anthropic IPO week · Nvidia petition and earnings · Mirendil, Mercor, Fractile funding

AI

How to connect Google Wallet to Gemini, and what it can see

Announced Nov 13 · US Gemini app users · web, Android and iOS · no extra cost mentioned

Google is rolling out a Google Wallet connection to every Gemini app user in the US. Once you turn it on, you can tag @Google Wallet in a chat and ask spending questions: chart last quarter's cash flow, find recurring charges that went up this year, or build a holiday budget from what you spent on gifts and travel last year.

Wallet data alone is thin, so Google lets you link credit card and bank accounts to Wallet through Plaid for fuller answers. That step is optional. Google says the connection is read-only: Gemini cannot move, manage or invest money, and the feature is opt-in.

How to turn it on: open Gemini's Personal Intelligence settings, go to Connected Apps and switch on Google Wallet. The same toggle turns it off. Anything Gemini saves about your goals shows up under Profile > Personal Intelligence > Memory Summary, where you can edit or delete it.

Our take: the subscription-price check is the most practical prompt, since quiet increases on recurring plans are easy to miss. If you link bank accounts, review the Memory Summary afterward.

AI API price changes and shutdowns from November 30 to December 11

Dates from Google, OpenAI and Anthropic developer docs, checked Nov 16

If you build on AI APIs, the next four weeks carry a cluster of deadlines. The biggest cost jump is Google's: Gemini 4 Carbon and Gemini 4 Argon double on December 1, from $2 to $4 per million input tokens and from $10 to $20 per million output tokens. We covered the Argon side in our Gemini 4 Argon price and access guide.

AI API deadlines to put on your calendar

DateVendorWhat changesReplacement or new price
Nov 30OpenAIGPT-5.6 Sol promo price ($4 in / $20 out) endsPost-promo price not published
Nov 30OpenAIAgent Builder, Evals and the Prompts API shut downAgents SDK or ChatGPT Workspace Agents; Promptfoo for evals
Nov 30Anthropicclaude-sonnet-4-5-20250929 retires (tentative)claude-sonnet-5-5
Dec 1GoogleGemini 4 Carbon and Argon: $2 / $10$4 / $20
Dec 1Googlegemini-2.5-pro calls rerouteGemini 3.8 Flash, billed $0.75 / $3.75
Dec 1OpenAIgpt-image-1-mini and gpt-image-1.5 shut downgpt-image-2.6-sunburst or -flare
Dec 11OpenAIgpt-5, gpt-5-mini, gpt-5-nano, o3 and o3-pro shut downgpt-6.1-sol, gpt-6.1-terra or gpt-6-luna
Jan 1, 2027GoogleGemini 3.8 Flash: $0.75 / $3.75$1.50 / $7.50
Prices are per million input / output tokens, standard paid tier. Sources: Google AI Studio pricing and deprecations pages, OpenAI deprecations page and GPT-5.6 Sol model page, Anthropic model deprecations page, all checked November 16, 2026.

What to do: search your code and agent configs for these model IDs now. Note the quiet trap in the Gemini 2.5 Pro reroute: your bill drops on December 1, then the Gemini 3.8 Flash rate it lands on doubles on January 1.

Cursor security breach update: who was affected and what to rotate

Update published Nov 13 · Cursor users, especially team admins

Cursor closed out a two-week investigation on November 13. The chain started when an attacker obtained an access key Cursor had issued to Fireworks AI, which Fireworks exposed. On October 29 the attacker planted malicious copies of the Claude Code extension in the cache behind Cursor's VS Code extension marketplace for about 11 hours.

Cursor incident: timeline and scope

Oct 29

Malicious Claude Code extension builds served for about 11 hours. At most 5,000 users could have received one.

Oct 30

Security notice with indicators of compromise and check scripts; notifications begin.

Nov 1

Last attacker activity Cursor observed, which was blocked.

Nov 13

Final update: under 0.01% of customers had API credentials exposed; agent traces exposed for under 0.05% of users outside Privacy Mode.

Percentages and counts are Cursor's own figures. Source: Cursor blog, "An update on our recent security incident," November 13, 2026.

Cursor says Privacy Mode code, production systems, the app and customer repositories were not touched; support emails, usage metrics and archived non-production code were.

What to do: if you installed or updated Anthropic.claude-code 2.1.312 or 2.1.313 from Cursor's marketplace on October 29, follow the October 29 notice; Cursor warns that uninstalling may not be enough. Team admins who got an email listing exposed keys should rotate them. No email means your data was not in the exposed traces. Our guide to vetting a Claude skill before you install it covers the same habits for add-ons.

Grok 4.8 release: limited rollout, no API or benchmarks yet

Rollout confirmed Nov 13 · Grok Bot users only

xAI's Grok account said on November 13 that a limited Grok 4.8 rollout to Grok Bot users was underway, according to Zeniteq. Three days earlier, Elon Musk said pretraining was done and that xAI had "rewritten the entire RL and inference stack from scratch in C/C++."

What is missing matters more than what shipped. As of November 15 there was no model card, no published benchmarks, no API model name and no rate card, Zeniteq and OrcaRouter report, so developers cannot call Grok 4.8 yet. Neither post says which Grok plans qualify. Musk also set expectations low, saying the model is "not yet incredible" and, asked whether it beats Claude Opus 5.5, answering "not as good," per OrcaRouter.

What to do: Grok Bot users can check the model picker; API buyers should wait for xAI's pricing.

Why did OpenAI disband its robotics team? Plus its Odessia travel buy

Robotics change reported Nov 13 · Odessia deal announced Nov 10

OpenAI confirmed to Fortune that it folded its standalone robotics team into other groups. Researchers now report to Chief Research Officer Mark Chen, and the applied robotics and hardware staff moved to the Applied organization under President Greg Brockman. The reason given: general models such as GPT-6 Astra can already run robotic tasks directly. A spokesperson called it a commitment rather than a retreat, and Sam Altman still targets a first public humanoid robot demo in 2027.

Separately, OpenAI bought Odessia, an AI travel agent that opened to the public on October 14 after raising about $6 million in a seed round led by Sequoia, Skift reports. Founder Francis Davidson joins OpenAI "to focus on the future of travel." Odessia's site now says the service "is no longer available."

What to do: Odessia customers with bookings should use the support email on its site. OpenAI hasn't said how the travel tech will appear in ChatGPT.

SaaS & business

Cloudflare Workers spending limit: Spend Management can now pause your bill

Beta launched Nov 13 · pay-as-you-go accounts · Workers and Durable Objects

Cloudflare's budget alerts used to warn you; the new Spend Management beta can stop the spending. Any pay-as-you-go customer can set a spending limit, alert thresholds and a notification email per product under Manage Account > Billing > Spend limits, or through the API and the cf CLI.

When a limit is hit, Cloudflare pauses the product: new billable Worker invocations are rejected, and Cron Triggers and Queue consumers stop, while static assets keep serving. To resume, raise the limit, remove the policy or wait for the next billing cycle. Caveats: spend data lags by up to 10 minutes, so a fast spike can pass the cap, and cycles start on your subscription date.

What to do: anyone running AI agents on Workers should set a cap today. Cloudflare plans to add more products and per-agent budgets later. For other vendors' 2026 plan moves, see our SaaS pricing changes tracker.

Anthropic IPO this week: pricing Wednesday, Nasdaq trading Thursday

Update to our Nov 15 coverage

Anthropic expects to price its shares on Wednesday, November 18, and start trading Thursday, November 19, on Nasdaq under ANT. The $70 to $78 range is unchanged, and with the private placement the raise could reach about $90.4 billion, Yahoo Finance reports. Dario Amodei told JPMorgan's Jamie Dimon the company wants to "underpromise and overdeliver."

New since our last update: TSMC's board approved up to NT$16.43 billion (roughly $500 million) for Anthropic shares in the private placement at the IPO price, and the same amount for OpenAI preferred shares, Taiwan News reports.

Anthropic's IPO by the numbers

ItemFigure
Shares offered825 million Class A at $70–$78
IPO proceeds$57.75B–$64.35B
Private placement at IPO price$26B of non-voting Class N stock; Google and Nvidia $10B each
Total, top of rangeUp to about $90.4B
Implied valuation, top of rangeAbout $1.404 trillion, fully diluted
TSMC's approved shareUp to NT$16.43B (about $500M)
Key datesPricing Wed, Nov 18 · Nasdaq trading Thu, Nov 19 (expected)
Sources: TechCrunch, November 10; Yahoo Finance, November 13; Taiwan News, November 11, 2026. Dates are company expectations and can move.

What it means for users: Claude plans and prices don't change because of the listing. Want shares? Ask your broker about IPO allocations. This is news, not investment advice.

Will Jensen Huang return to Nvidia? Petition tops 115,000 signatures

Petition counted Nov 14 · earnings Nov 17

A petition at standwithjensen.com asking Nvidia to reinstate Jensen Huang passed 115,000 signatures by Saturday afternoon, TechCrunch reports, including 5,965 people who identify as current Nvidia employees and 63,063 who identify as shareholders. Elon Musk, Replit CEO Amjad Masad, Epic Games CEO Tim Sweeney and Y Combinator CEO Garry Tan signed. Altimeter Capital and Atreides Management, holding about $2.1 billion in Nvidia stock, asked the board to keep Huang's return open during a slow CEO search, CNBC reported.

Nvidia has not said Huang will return. Interim CEO Colette Kress leads Tuesday's results, due about 1:20 p.m. PT, with the call at 2 p.m. PT.

Why software buyers should care: Nvidia's guidance shapes GPU supply and cloud pricing, which flows into what AI tools charge. Watch Tuesday's data center outlook more than the leadership drama.

AI startup funding this week: Mirendil, Mercor and Fractile raise $2.47 billion

Rounds announced Nov 10–12

Three big rounds we had not covered yet bet on research automation, expert training data and inference chips.

Largest AI rounds we tracked, Nov 10–12 (US$ millions)

Mirendil · Series A · valuation not disclosed

$1,000M

Mercor · Series D · $20B valuation

$750M

Fractile · Series C · $7B valuation

$720M
Sources: Mirendil (Nov 12), Mercor (Nov 10) and SiliconANGLE on Fractile (Nov 11), 2026. Mercor's round includes an employee tender.
  • Mirendil, backed by Kleiner Perkins and Clearlake Capital with Google participating, makes Lor, a private-beta tool that runs hundreds of AI agents on code and experiments for days.
  • Mercor pays a network of more than five million experts to train and evaluate AI models; Nvidia, Salesforce and Zoom joined the round.
  • Fractile, led by Thrive Capital, Founders Fund, Lightspeed and Redpoint, stacks DRAM on top of its inference chips and plans to ship them in 2027.

Questions readers are asking this Monday

Can Gemini move money out of my bank account?

No. Google says the Wallet connection is read-only. Gemini can read Wallet data and any bank or card accounts you link through Plaid to answer spending questions, but it cannot move, manage or invest money. You can switch the connection off anytime under Connected Apps in Personal Intelligence settings.

I use Cursor. Do I need to do anything?

Only in two cases. If you installed Claude Code extension version 2.1.312 or 2.1.313 from Cursor's marketplace on October 29, follow Cursor's October 29 security notice, because uninstalling may not be enough. If your admin got an email listing exposed keys, rotate those keys. No notification means your data wasn't exposed.

When does Anthropic stock start trading?

Anthropic expects to price its IPO on Wednesday, November 18, and begin trading on Thursday, November 19, on Nasdaq under the ticker ANT. The price range is $70 to $78 a share. Dates can still move, and whether you can buy at the IPO price depends on your broker.

How much will Gemini 4 Carbon cost after December 1?

On Google's standard paid tier, Gemini 4 Carbon goes from $2 to $4 per million input tokens and from $10 to $20 per million output tokens on December 1, 2026. Gemini 4 Argon doubles the same way. Gemini 4 Hydrogen has no change listed and stays at $1.50 input and $7.50 output.

Sources: Google, Nov 13 · Google AI Studio pricing, checked Nov 16 · Google deprecations, checked Nov 16 · OpenAI deprecations, checked Nov 16 · Anthropic deprecations, checked Nov 16 · Cursor, Nov 13 · Zeniteq, Nov 15 · OrcaRouter, Nov 10 · Fortune, Nov 13 · Skift, Nov 10 · Cloudflare, Nov 13 · TechCrunch, Nov 10 · Yahoo Finance, Nov 13 · Taiwan News, Nov 11 · TechCrunch, Nov 14 · CNBC, Nov 14 · Nvidia · Mirendil, Nov 12 · Mercor, Nov 10 · SiliconANGLE, Nov 11

Tags: #ai saas news today
Quan Le
ABOUT THE AUTHOR

Quan Le

Quan Le is the founder and editor of Saazora, an independent site for news and reviews of SaaS and AI tools. He has worked in affiliate publishing since 2015, writing product reviews, verifying deals and running SEO for review sites. On Saazora he focuses on software for small businesses, creators and marketing teams, and evaluates tools by how well they fit real workflows: onboarding, integrations, pricing and long-term value, not the length of their feature list.

Read Quan's full profile · LinkedIn · contact@saazora.com

KEEP READING

Related Reviews