HomeInsightsSignal / NoiseSignal / Noise Nov 19
Signal / Noise

Signal / Noise (Nov 19, 2026): Microsoft To Do Retirement, Gemini 4 Neon Price, OpenAI End Dates, Amazon Connect Price

WWaveo teamNovember 19, 202610 min read
Five glass tiles in a ring on a round metal platform, a thin red beam crossing the first three, which glow red, and ending at a glowing node in the center, while the other two stay clear and dim.
The short version

Microsoft says To Do apps retire on June 30, 2027 and lists will not move to Planner. Google prints $3 and $15 per million input and output tokens for Gemini 4 Neon. Amazon prints $0.40 per case (one incoming request), with 4 steps included, for Resolutions. OpenAI lists end dates for models and service tiers from January 11, 2027. One thing to do: find out who uses To Do.

In plain words

Think of this edition as a wall clock, a price tag and a shutdown list, plus a short note. The clock is Microsoft's date for switching off its To Do apps. The price tags are what Google and Amazon print for their AI services: Google bills by the token (a chunk of text), Amazon by the case. The list is OpenAI's end dates. The short note is about one company's public notice on selling its shares.

Four Signal items and one short Watch item. Signal means you can act on it this week; Watch means worth knowing. Every number belongs to whoever published it; the "What it means for you" and "Our take" lines are ours. Previous: November 18. Also published today: our decode of a Dutch regulator's page on AI camera glasses and our teardown of Stripe Treasury for agents.

Short on time? Read the Microsoft item and the closing box: two minutes. Read the Gemini item only if you buy AI by the token, the OpenAI table only if a tool of yours names an OpenAI model, the Amazon item only if you run a contact center. The last item needs no action.

Which dates matter in this edition?

DateWhat happens
Jan 11, Feb 10, Feb 26 and Apr 1, 2027OpenAI shutdown dates (item 4); January 11 is for EU fast and priority tiers
June 30, 2027Microsoft: To Do apps retire

Signal: When do Microsoft's To Do apps retire, and what happens to your lists?

Microsoft To Do is Microsoft's app for task lists (our gloss). Microsoft's support page says "Microsoft To Do apps will retire on June 30, 2027." That covers the web, Windows, Mac, iOS and Android apps, "for both work or school accounts and personal Microsoft accounts". The Planner blog post of November 18, 2026 repeats the date. The support page's text prints no date, so the blog post is the dated source.

What happens to your lists?

Short answer: your lists will not move by themselves.

Work or school accounts. The support page says "Existing To Do tasks and lists will not migrate to Planner." (Planner is Microsoft's task and project app, our gloss.) Existing tasks "will remain accessible in Planner through June 30, 2027", after which you can no longer view or manage them there. The page does not scope that sentence. In our reading it applies to work or school accounts, because the page says Planner is available for them and not for personal accounts.

Personal accounts. The page says "Planner is not available for these accounts." and tells people to "Choose another task-management application that meets your needs."

Your data. "App retirement does not itself schedule existing task data for deletion", the page adds. The pages we read do not state how long To Do data is kept.

Export. Microsoft says it will provide "a way to export your To Do data" and will add instructions "when those details are finalized"; an export "does not migrate tasks or lists into Planner". The pages we read (the support page and the Planner blog post) do not yet publish export steps, a date for the To Do service and API retirement ("separate, later timelines"), or a price for Planner.

Time left. By our arithmetic, June 30, 2027 is 223 days after November 19, 2026, just over seven months.

Classic Outlook Tasks. The support page says classic Outlook's native Tasks experience is separate from the To Do experience hosted in Outlook and "will continue working and syncing with Exchange".

What it means for you: in our reading, find out who in your company still uses To Do, on a work or school account or a personal one, and which shared lists matter. Ask your admin to confirm Planner access and licenses, and watch the support page for the export steps.

Our take: Signal, because it is a fixed date with a short list of questions. Status as of November 19, 2026.

Signal: What does Google print for Gemini 4 Neon's price?

Gemini 4 Neon is an AI model that Google introduced on November 18, 2026 (our gloss). Developers pay to use models like it per token, a chunk of text that is often part of a word (our gloss). Google's blog says it is "priced at $3 per million input tokens" and "$15 per million output tokens", with a 1 million-token context window (how much text it can take in at once; our gloss), and is "rolling out today" in the Gemini API in Google AI Studio and in other Google products.

We read Google's pricing page in a browser. On its Standard tab, the paid-tier column lists, per 1 million tokens in US dollars:

Line on the pagePrice
Input price$3.00
Output price (including thinking tokens)$15.00
Context caching price$0.15, with the note "Implicit caching only (explicit caching is not supported)."

The page says "Gemini 4 Neon is not available on the Free tier." It also says every token the model processes is counted and billed, including thinking tokens (text the model writes while it works on a problem, our gloss).

By our arithmetic, the output price is five times the input price, and 1 million input plus 1 million output tokens come to $3.00 + $15.00 = $18.00. That is a price-list example, not a bill forecast; a job's cost depends on its tokens (our reading). Google's price step for a different Gemini model on December 1 is in our November 11 edition.

The pages we read (the blog and the Standard tab) do not state a dollar cost for a typical job, and the pricing page prints no date.

What it means for you: in our reading, if your team or a vendor you pay calls the Gemini API, ask which Gemini model it uses and what the monthly token bill is.

Our take: Signal, because it is a printed price you can set against your own bill. Status as of November 19, 2026.

Signal: How does AWS price Resolutions in Amazon Connect Customer?

In plain words, Resolutions handles routine back-office requests step by step with AI agents and human specialists (our gloss). AWS's headline calls it "an agentic solution for high-volume operational work". AWS's What's New post of November 19, 2026 "announces the general availability of Resolutions" in Amazon Connect Customer and says it "opens a Resolutions Case for each incoming request".

What you are billed forWhat AWS's pricing appendix prints
Each Resolutions Case created"$0.40 per Resolutions Case created, including 4 steps"
Each block of 5 additional steps$0.40 per block; "A partial block is billed as a full block of 5 steps."
Customer communications through Connect channels (voice, chat, email, SMS)Not part of Resolutions pricing; billed at standard channel rates

The appendix says a step is billed whether a human specialist, an AI agent or a call to a business application performs it, and gives "such as an approval or exception review" as a human example. Retries caused by system issues are not additional steps; rework you request is. AWS's own example: a case with 5 steps costs $0.40 plus $0.40, $0.80 in total.

Steps in one caseTotal (our arithmetic)
1 to 4$0.40
5 to 9$0.80
10 to 14$1.20

AWS prints only the 5-step example; channel charges are extra. In our reading, a human approval is a billed step like any other: if a case already has four steps, an approval as the fifth moves it from $0.40 to $0.80.

The pages we read (the What's New post and the pricing appendix) do not state, for Resolutions, a volume discount, a cap, a measured result, a named customer, or what counts as one step beyond the examples. We did not read the documentation or the product page, which the post links, or the main Amazon Connect pricing page.

What it means for you: in our reading, if you run or compare a contact center or back-office process, count the steps in one typical request before you read a price.

Our take: Signal, because the rule is printed in full and one line of it (a human step is billed) shapes how a process is built. Status as of November 19, 2026.

Signal: Some OpenAI end dates, from January 11 to April 1, 2027

OpenAI's deprecations page is its public list of models and features it is switching off (our gloss). It lists many shutdown dates; the rows below are the ones from January 11 to April 1, 2027 that our earlier editions did not cover. Its legend says "Deprecated" means "You can still use it until its shutdown date." Every row below has that status.

The table covers models and service tiers and groups rows that share a shutdown date, a replacement and a notice date; this table does not include one row dated February 1, 2027 for a specialist model; check the page for it. Replacements are as OpenAI names them. Only the January 11 row has a notice date of November 18. The page does not say when rows were added. Our November 18 edition already noted the later February and April dates.

Shutdown dateAffectedRecommended replacementNotice date
Jan 11, 2027GPT-5.2, GPT-5.4, and GPT-5.4 mini with EU data residency (data kept in the EU, our gloss) and service_tier: fast or priority (a row in the page's feature and endpoint table)The same models with EU data residency and Standard processing (service_tier: default), or the GPT-6.2 model familyNov 18, 2026
Feb 10, 2027gpt-image-2; gpt-image-2-2026-04-21gpt-image-2.6-sunburst or gpt-image-2.6-flareNov 10, 2026
Feb 26, 2027whisper-1; gpt-4o-transcribe; gpt-4o-mini-transcribe and gpt-4o-mini-transcribe-2025-12-15; gpt-4o-transcribe-diarizegpt-live-transcribe or gpt-transcribeAug 26, 2026
Feb 26, 2027/v1/audio/translations (an endpoint)/v1/realtime/translationsAug 26, 2026
Apr 1, 2027gpt-5.3-codex; gpt-5.1 and gpt-5.1-2025-11-13gpt-6-solOct 1, 2026
Apr 1, 2027gpt-5.4-nano and gpt-5.4-nano-2026-03-17gpt-6-lunaOct 1, 2026

From November 19, January 11 is 53 days away, February 10 is 83, February 26 is 99 and April 1 is 133 (our arithmetic).

Rows we covered before are in our November 17 edition and our November 18 edition. These dates now sit across three of our editions; we may put them on a single page, an idea, not a promise.

What it means for you: in our reading, ask whoever built your AI tools for one list of every model and tool they use, with end dates. We asked for the same list in our November 18 edition. Add these dates to it and check which tool names a model in the table.

Our take: Signal, because the vendor fixed the dates and the first step is one list. Status as of November 19, 2026: every row shown is "Deprecated".

Watch: What does Anthropic's own notice say about its share offering?

Anthropic's announcement of November 18, 2026 about a first public sale of its shares (our gloss) says it "priced its initial public offering of its Class A common stock at $83.00 per share". It lists 980,000,000 shares, with trading expected from November 19 and closing expected November 20, 2026, subject to customary closing conditions. We did not check whether either has happened. The page prints no total raised, valuation, first trading price, customers or contracts. We read no prospectus or SEC filing. Not investment advice. Status as of November 19, 2026.

If you do one thing this week

In our reading, start with the first line; the other two are quick checks. Not legal or IT advice.

  1. By Friday, November 20, find out who in your company still uses Microsoft To Do, on a work or school account or a personal one, and put June 30, 2027 on the calendar (223 days after November 19, our arithmetic).
  2. Add today's OpenAI dates to the one list we asked for on November 18: every model and tool your builders use, with its end date. The first date, January 11, is 53 days from November 19 (our arithmetic).
  3. If you or a vendor you pay calls the Gemini API, ask which model it uses.
Book an AI discovery call

How we know, and what we could not check

We read the primary pages on November 19, 2026, from about 14:49 PST, and the Gemini pricing page in a browser at about 14:52 PST. We tested nothing. Not read: Microsoft's Planner admin pages; Google's other pricing tabs; AWS's documentation, product page and main Amazon Connect pricing page; OpenAI's pricing pages; press; and any prospectus or SEC filing (an automated read of the SEC's filing site was refused on November 19 and we did not work around it). We are not lawyers or financial advisers; this describes documents, not a rule or a recommendation.

Sources

FAQ

When do Microsoft's To Do apps retire?

Microsoft's support page says "Microsoft To Do apps will retire on June 30, 2027." It covers the web, Windows, Mac, iOS and Android apps for work or school accounts and personal Microsoft accounts. Microsoft's Planner blog post of November 18, 2026 repeats the date; the support page's text prints no date. By our arithmetic, June 30, 2027 is 223 days after November 19, 2026.

Will my Microsoft To Do lists move to Planner?

The support page says existing To Do tasks and lists will not migrate to Planner. It says existing tasks will remain accessible in Planner through June 30, 2027 (Planner is available for work or school accounts, our reading), after which you can no longer view or manage them there, and app retirement does not itself schedule existing task data for deletion. For personal Microsoft accounts the page says Planner is not available. Microsoft says it will provide a way to export To Do data; the pages we read do not yet publish export steps.

How much does Gemini 4 Neon cost?

Google's blog of November 18, 2026 lists $3 and $15 per million input and output tokens (and $0.15 per million cached input tokens). The Standard tab of its pricing page lists $3.00, $15.00 (including thinking tokens) and $0.15 for context caching. These are Google's prices; we did not read the Batch, Flex or Priority tabs. Per-token prices do not say what one job costs, which depends on how many tokens it uses.

Is Gemini 4 Neon offered on Google's entry tier for the API?

Google's pricing page says "Gemini 4 Neon is not available on the Free tier." The prices on the page are for the paid tier. We read the Standard tab only.

Which OpenAI models and features end in early 2027?

OpenAI's deprecations page lists a January 11, 2027 shutdown date for GPT-5.2, GPT-5.4 and GPT-5.4 mini with EU data residency and service tier fast or priority (notice date November 18, 2026), and later shutdown dates of February 10, February 26 and April 1, 2027 with earlier notice dates. This answer does not include a row dated February 1, 2027 for a specialist model. The page does not say when rows were added. Earlier OpenAI dates are in our November 17 and November 18 editions.

What does an Amazon Connect Resolutions case cost?

AWS's pricing appendix lists $0.40 per Resolutions Case created, including 4 steps, then $0.40 per block of 5 additional steps; a partial block is billed as a full block. A step is billed whether a human specialist, an AI agent or a call to a business application performs it. Customer communications through Connect channels are billed separately. We did not read the documentation, the product page or the main Amazon Connect pricing page.

Which dates in this edition should I put on a calendar?

From this edition: June 30, 2027 for the Microsoft To Do apps, and January 11, February 10, February 26 and April 1, 2027 for the OpenAI rows. Microsoft also says the updated My tasks experience in Planner begins rolling out in December 2026. Earlier OpenAI dates are in our November 17 and November 18 editions.

Work with us

Not sure which of your tools depend on To Do or an OpenAI model?

Tell us which AI and productivity tools your team uses and we will help you map which vendors, models and dates each one depends on.

Book an AI discovery call