Hugin / Deadlines
The clock.
Every end date a provider has published, counting down.
24 upcoming32 passedSubscribe by calendar
- 8 days
Amazon Web Services · Retirement
Five older Claude model families in the Regions specified in Amazon Bedrock's lifecycle table reach their scheduled end of life
Check your exact model ID and AWS Region against the operative table. Amazon says migration does not happen automatically; plan a transition to an Active model before the listed EOL date.
Nothing to move to. The captured legacy table has no replacement column and names no successor for these rows. This does not establish that no alternative exists.
Sources and checks
Read October 22, 2026. Region-scoped rows, not worldwide vendor retirements. The page's general prose promises at least six months in Legacy and at least three months before extended access, but these rows list October 10/13/15 Legacy starts, same-day extended access and November 13 EOL. That source inconsistency is unresolved; the table dates are reported, not reconciled. Amazon says requests fail on or soon after EOL unless privately arranged. No authenticated endpoint was called. The exact table rows below retain the date, model IDs and Region scope; continuation rows inherit the model named immediately before them. Anthropic Claude 3 Haiku anthropic.claude-3-haiku-20240307-v1:0 ap-northeast-2, ap-south-1, ap-southeast-1 October 10, 2026 November 13, 2026 October 10, 2026 | eu-central-2, sa-east-1 October 13, 2026 November 13, 2026 October 13, 2026 | ca-central-1 October 15, 2026 November 13, 2026 October 15, 2026 | Anthropic Claude 3 Sonnet anthropic.claude-3-sonnet-20240229-v1:0 ap-south-1, ap-southeast-1, eu-west-2 October 10, 2026 November 13, 2026 October 10, 2026 | sa-east-1 October 13, 2026 November 13, 2026 October 13, 2026 | ca-central-1 October 15, 2026 November 13, 2026 October 15, 2026 | Anthropic Claude 3.5 Sonnet anthropic.claude-3-5-sonnet-20240620-v1:0 ap-northeast-2, ap-south-1, ap-southeast-1 October 10, 2026 November 13, 2026 October 10, 2026 | Anthropic Claude 3.5 Sonnet v2 anthropic.claude-3-5-sonnet-20241022-v2:0 ap-northeast-2, ap-south-1, ap-southeast-1 October 10, 2026 November 13, 2026 October 10, 2026 | Anthropic Claude 3.7 Sonnet anthropic.claude-3-7-sonnet-20250219-v1:0 ap-south-1, eu-west-2 October 10, 2026 November 13, 2026 October 10, 2026
Operative page re-read: Read, and it still says what this row says: 28 recorded terms present.
- 12 days
Google · Retirement
Five Gemini API preview models reach their listed shutdown date: the speech previews gemini-2.5-flash-preview-tts, gemini-2.5-pro-preview-tts and gemini-3.1-flash-tts-preview, and the live previews gemini-3.1-flash-live-preview and gemini-2.5-flash-native-audio-preview-12-2025
Search your configuration for the five exact model ids and repoint them: the three TTS ids at gemini-3.8-flash-tts or gemini-3.8-flash-lite-tts, the two live ids at gemini-3.8-live. All five are preview endpoints, which Google's deprecation policy treats differently from generally available models — the replacements are generally available ids, not previews.
Moves to: From the table's replacement column: gemini-3.8-flash-tts or gemini-3.8-flash-lite-tts for all three speech previews, and gemini-3.8-live for both live previews.
Sources and checks
One date, five ids, and only on the Gemini API surface. The table states a listed shutdown date is the earliest on which a model might be retired, so November 17 is a floor; Google says it will communicate the exact date with advance notice. Replacing a speech or live-audio preview is not always a find-and-replace — voice, latency and streaming behaviour can differ between a preview and its generally available successor, and the table says nothing about that.
Operative page re-read: Read, and it still says what this row says: 24 recorded terms present.
- 14 days
Google · Retirement
gemini-3.6-flash retires on Google Cloud's Gemini Enterprise Agent Platform, four months after its release
If an Agent Platform call names gemini-3.6-flash, move it to gemini-3.8-flash before November 19. This is a short-term availability model: the page says such a model stays active until a replacement ships and a date is announced, so the date is new rather than moved, and the announcement is the warning.
Moves to: gemini-3.8-flash, named in the short-term availability table's replacement column.
Sources and checks
Google Cloud's schedule only; the Gemini API's deprecations page is a separate clock. The page commits to at least 45 days' notice for a short-term model and says a posted retirement date can be extended but not pulled earlier, so this is a floor. A model released July 21, 2026 and retiring November 19, 2026 is a four-month life by this desk's count — short, and exactly what the short-term table is for.
Operative page re-read: Read, and it still says what this row says: 17 recorded terms present.
- 25 days
OpenAI · Retirement
The v1/prompts API and reusable prompt objects shut down
Move reusable prompt content into your application code — the ledger's stated migration is inlining, not a replacement API.
Moves to: No replacement API. The page's stated migration is to "move reusable prompt content into your application code", with a linked Migrate from prompt objects guide — a path, and not a product.
Sources and checks
Announced June 3, 2026, with prompt creation de-emphasized in the platform the same day. Lands on the same date as the Agent Builder and Evals shutdowns — three products end on November 30.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 25 days
OpenAI · Retirement
The Evals dashboard and API shut down
Migrate eval workflows before the date; the ledger names Promptfoo as the documented path. Evals are already read-only from October 31.
Moves to: Promptfoo, a third-party tool: the page links "Moving from OpenAI Evals to Promptfoo for a migration path". Worth reading twice — the continuation is off-platform.
Sources and checks
The second Evals step — the October 31 read-only cliff is its own row. Graders documented for eval workflows are stated to be part of this transition; fine-tuning timelines are separate.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 25 days
OpenAI · Retirement
Agent Builder shuts down
Migrate to the Agents SDK or ChatGPT Workspace Agents — both are named on the page as the continuation path.
Moves to: The Agents SDK or ChatGPT Workspace Agents, via a linked Migrate from Agent Builder guide; ChatKit is stated to remain available.
Sources and checks
The ledger states that ChatKit remains available. The shutdown is of the builder product, not of every agent surface.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 26 days
Google · Price change
Every per-token Gemini API price on gemini-4-argon doubles: Standard input from $2.00 to $4.00 and output from $10.00 to $20.00 per million tokens, with Batch, Flex, Priority, the invite-only Contributor tier and context caching stepping the same way
If you run Google's flagship reasoning model, budget December at twice your November spend on it — a month before the Gemini 3 Flash family's own doubling on January 1, 2027. A spend cap or prepaid balance set in dollars is reached twice as fast by unchanged traffic, so check both before December 1.
Sources and checks
Listed prices on Google's pricing page, which reads '$2.00 through November 30, 2026.' then '$4.00 starting December 1, 2026.' for Argon's Standard input; not a customer notice and not a bill. Every per-token rate on the model steps to exactly twice its current figure; the grounding charges carry no dated second figure. Argon has no free tier to be unaffected. The step is a month earlier than the Flash family's January 1, 2027 step recorded in google-gemini-flash-paid-price-doubling, and Google's changelog names no Argon figure — the same pattern the September 28 record found on the Flash models. Gemini Developer API only: Google Cloud's Agent Platform prices were not read, and this desk holds no invoice.
Operative page changed: Read, and something changed: 1 recorded term no longer on the page. Open the page before relying on this row.
- 26 days
OpenAI · Retirement
The older GPT Image models — gpt-image-1-mini, gpt-image-1.5 and the chatgpt-image-latest alias — are removed from the API
Repoint image calls at gpt-image-2, the named replacement for all three.
Moves to: gpt-image-2, named for all three ids.
Sources and checks
chatgpt-image-latest is a floating alias, so nothing in your code names a version that could warn you — the id keeps working right up to the date and then does not. Notified June 2, 2026.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 36 days
OpenAI · Retirement
The first-generation GPT-5 snapshots — gpt-5, gpt-5-mini, gpt-5-nano and gpt-5-pro — and the o3 and o3-pro reasoning models are removed from the API
Repoint at the named substitutes: gpt-5.6-sol for gpt-5, gpt-5-pro, o3 and o3-pro (pro via reasoning.mode), gpt-5.6-terra for gpt-5-mini, gpt-5.6-luna for gpt-5-nano.
Moves to: gpt-5.6-sol, -terra or -luna per snapshot; the two pro models map to gpt-5.6-sol with reasoning.mode: pro.
Sources and checks
Announced June 11, 2026. Sixteen months after GPT-5's release, its launch snapshots retire — every id in this group is snapshot-dated, so a pinned snapshot fails on the date while a floating alias will already have moved.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 36 days
OpenAI · Retirement
Custom GPTs retire across ChatGPT plans — GPTs and their GPT pages become inaccessible
If you built a custom GPT, migrate it to a plugin and test it before December 11, and rebuild any custom actions — they do not transfer. If you rely on someone else's GPT, check for the creator's replacement; permission to use a GPT does not give you permission to migrate it.
Moves to: Plugins: the FAQ's migration turns a GPT's instructions into a skill within a new plugin and copies its knowledge files. Custom actions, past conversations, sharing settings and the selected model do not carry over.
Sources and checks
Scheduled, not executed. 'The transition affects all ChatGPT plans', but Enterprise workspaces with an approved deferral retire on February 11, 2027 instead, and the FAQ tells every reader to follow the notice for their own account or workspace. The FAQ is served by a host that refuses this desk's client intermittently.
Operative page re-read: Read, and it still says what this row says: 10 recorded terms present.
- 57 days
Google · Price change
Paid Gemini API prices double for Gemini 3.6 Flash, 3.7 Flash and 3.8 Flash: standard input from $0.75 to $1.50 and output from $3.75 to $7.50 per million tokens, with the 3.8 Flash TTS and 3.8 Flash-Lite TTS speech models doubling the same day
Budget January at twice your current spend on these models. A spend cap or prepaid balance set in dollars will be reached twice as fast by the same traffic, so check both before the date.
Sources and checks
Listed prices on Google's pricing page, which reads '$0.75 through December 31, 2026.' then '$1.50 starting January 1, 2027.' for all three models' standard input; not a customer notice and not a bill. Batch, Flex, Priority and context caching step the same way, the free tier stays 'Free of charge', and Gemini 3.5 Flash has no step. Google's changelog calls the 3.7 Flash rate introductory without naming a figure and mentions no step for the others. Gemini API only: prices on Google Cloud's Agent Platform were not read.
Operative page changed: Read, and something changed: 1 recorded term no longer on the page. Open the page before relying on this row.
- 62 days
OpenAI · Rule due
Active existing customers can no longer create new self-serve fine-tuning jobs
If a fine-tuned model is part of your roadmap, the job has to exist before this date — creation ends, not inference.
Sources and checks
Scoped to CREATING jobs. The ledger states that inference on already fine-tuned models is disabled only when the underlying base model is deprecated, which is a separate date per base model.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 62 days
OpenAI · Retirement
Four text-to-speech models are removed from the API — tts-1, tts-1-hd, gpt-4o-mini-tts-2025-03-20 and gpt-4o-mini-tts-2025-12-15
If you synthesise speech through tts-1, tts-1-hd or either dated gpt-4o-mini-tts snapshot, repoint at gpt-realtime-2.1-mini and retest before January 6. The section's own instruction is to 'See the Realtime API guide to plan your migration', so treat it as a migration to plan rather than a model id to swap.
Moves to: gpt-realtime-2.1-mini, named for all four ids and in the section's own migration sentence.
Sources and checks
Announced in a ledger section dated October 1, 2026, which states the removal comes 'with at least three months’ notice' — a floor rather than a figure, and the shorter of the two notice periods stated by the two sections the ledger dates October 1. tts-1 and tts-1-hd are undated family ids widely embedded as defaults rather than choices, so this is more likely to be found in a dependency than in a config. The named replacement is a realtime model rather than a speech endpoint; the ledger names it without stating the two produce comparable output, and this desk holds no OpenAI key and has not compared them.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 70 days
OpenAI · Retirement
Five dated GPT-4 series snapshots are removed from the API — gpt-4o-2024-08-06, gpt-4o-2024-11-20, gpt-4o-mini-2024-07-18, gpt-4.1-2025-04-14 and gpt-4.1-mini-2025-04-14
Search your configuration for those five exact snapshot ids and for any fine-tuned model built on them; the section states that 'Fine-tuned versions of these models follow the same shutdown schedule as their underlying base models.' The ledger names gpt-6-luna as the replacement for all of them.
Moves to: gpt-6-luna — the ledger's recommended replacement column names it for every one of the section's four rows.
Sources and checks
Announced in a ledger section dated October 14, 2026, which says developers were 'notified' that day and — unlike the two sections dated October 1 immediately below it — states NO notice period, so there is no figure here for this desk to pass on. Every id in this group is a dated snapshot rather than a floating alias, so code naming one is readable in a diff; the exposure is instead in fine-tuned models, which the section sweeps in by reference to their base models without listing a single one. All four rows name the same replacement, gpt-6-luna, for four differently sized models across two families; the ledger names a replacement and does not say it is equivalent to what it replaces, and this desk holds no OpenAI key and has not compared them. The ledger records a schedule and keeps carrying ids after their dates, so it cannot by itself confirm a shutdown happened.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 76 days
OpenAI · Retirement
Every legacy audio, realtime and transcription family is removed from the API — gpt-realtime, gpt-audio, gpt-4o-audio, gpt-4o-realtime, their mini variants, and gpt-4o-mini-transcribe
Repoint at the named replacements: gpt-realtime-2.1 and gpt-realtime-2.1-mini for realtime, gpt-audio-1.5 for every audio family, gpt-4o-mini-transcribe-2025-12-15 for transcription.
Moves to: gpt-realtime-2.1 and gpt-realtime-2.1-mini for realtime, gpt-audio-1.5 for the audio families, gpt-4o-mini-transcribe-2025-12-15 for transcription.
Sources and checks
Announced July 20, 2026. This group retires whole FAMILIES, not dated snapshots — a floating family id like gpt-realtime keeps working right up to the date with nothing in your code naming a version that could warn you, the chatgpt-image-latest trap at family scale.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 84 days
Google · Retirement
gemini-3.7-flash retires on Google Cloud's Gemini Enterprise Agent Platform
If an Agent Platform call names gemini-3.7-flash, move it to gemini-3.8-flash before January 28, 2027. Both of the flash models this table retires point at the same successor, so a caller on either can make the move once.
Moves to: gemini-3.8-flash, named in the short-term availability table's replacement column — the same successor the table names for gemini-3.6-flash.
Sources and checks
Google Cloud's schedule only. The page commits to at least 45 days' notice for a short-term availability model and says a posted date can be extended but not pulled earlier. gemini-3.8-flash itself appears in the same table with no retirement date announced, which under the table's own rule means it keeps answering until Google posts one.
Operative page re-read: Read, and it still says what this row says: 17 recorded terms present.
- 113 days
OpenAI · Retirement
The legacy transcription models are removed from the API — whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe and gpt-4o-transcribe-diarize
Repoint at gpt-live-transcribe or gpt-transcribe, the two replacements the ledger names for all four ids.
Moves to: gpt-live-transcribe or gpt-transcribe, named for all four ids.
Sources and checks
Announced August 26, 2026, and the farthest date on the ledger — which is exactly why it is the one to write down. whisper-1 is embedded in a great deal of shipped code as a default rather than a choice, so this is more likely to be found in a dependency than in a config.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 124 days
Amazon Web Services · Retirement
Writer Palmyra X4 reaches scheduled end of life in its listed Bedrock Regions
Check your exact model ID and AWS Region against the operative table. Amazon says migration does not happen automatically; plan a transition to an Active model before the listed EOL date.
Nothing to move to. The captured legacy table has no replacement column and names no successor for these rows. This does not establish that no alternative exists.
Sources and checks
Read October 22, 2026. Region-scoped rows, not worldwide vendor retirements. Amazon says requests fail on or soon after EOL unless privately arranged. No authenticated endpoint was called. The exact table rows below retain the date, model IDs and Region scope; continuation rows inherit the model named immediately before them. Writer Palmyra X4 writer.palmyra-x4-v1:0 us-east-1, us-east-2, us-west-1, us-west-2 September 9, 2026 March 9, 2027 —
Operative page re-read: Read, and it still says what this row says: 28 recorded terms present.
- 130 days
Google · Retirement
gemini-2.5-flash-image reaches its earliest listed shutdown date in the Gemini API
Repoint calls that name gemini-2.5-flash-image at gemini-3.1-flash-lite-image before March 15, 2027. Google lists that as the earliest possible shutdown, so the model may keep answering past it — plan for the date, do not wait for a failure. Google Cloud's Agent Platform now gives the same id the same date, so a caller on either surface has one deadline rather than two.
Moves to: gemini-3.1-flash-lite-image, named in the deprecations table's replacement column as of 2026-10-07 — the same replacement Google Cloud's Agent Platform names for this id. Until a date between 2026-09-27 and 2026-10-07 the table named gemini-3.1-flash-image-preview instead, a preview listed with its own June 25, 2026 shutdown.
Sources and checks
Google's table states listed shutdown dates are 'the earliest possible dates on which a model might be retired' — a floor, not a schedule — and that it will give advance notice of the exact date. The move from October 2, 2026 to March 15, 2027 is this desk's reading of two dated fetches of the same page, on September 27 and October 7; Google published no release note this desk found announcing it. On September 14 the page sent an unauthenticated plain client to a sign-in flow and was read through a rendering fetcher; on September 27 and October 7 it answered curl.
Operative page re-read: Read, and it still says what this row says: 24 recorded terms present.
- 130 days
Google · Retirement
gemini-2.5-flash-image retires on Google Cloud's Gemini Enterprise Agent Platform, after an extension from October 2, 2026
On Google Cloud, plan the move to gemini-3.1-flash-lite-image before March 15, 2027. On the Gemini API the same model id has an October 2, 2026 earliest shutdown instead — check which surface you call.
Moves to: gemini-3.1-flash-lite-image, named in the September 14 release note and the model-versions table — a different replacement from the one the Gemini API names for the same id.
Sources and checks
Extended once already, on September 14, from October 2, 2026. Google says Cloud retirement dates may be extended but will not be moved earlier than listed.
Operative page re-read: Read, and it still says what this row says: 17 recorded terms present.
- 136 days
Amazon Web Services · Retirement
Writer Palmyra X5 reaches scheduled end of life in its listed Bedrock Regions
Check your exact model ID and AWS Region against the operative table. Amazon says migration does not happen automatically; plan a transition to an Active model before the listed EOL date.
Nothing to move to. The captured legacy table has no replacement column and names no successor for these rows. This does not establish that no alternative exists.
Sources and checks
Read October 22, 2026. Region-scoped rows, not worldwide vendor retirements. Amazon says requests fail on or soon after EOL unless privately arranged. No authenticated endpoint was called. The exact table rows below retain the date, model IDs and Region scope; continuation rows inherit the model named immediately before them. Writer Palmyra X5 writer.palmyra-x5-v1:0 us-east-1, us-east-2, us-west-1, us-west-2 September 21, 2026 March 21, 2027 —
Operative page re-read: Read, and it still says what this row says: 28 recorded terms present.
- 137 days
Amazon Web Services · Retirement
TwelveLabs Marengo Embed 3.0 and Pegasus v1.2 reach scheduled end of life in their listed Bedrock Regions
Check your exact model ID and AWS Region against the operative table. Amazon says migration does not happen automatically; plan a transition to an Active model before the listed EOL date.
Nothing to move to. The captured legacy table has no replacement column and names no successor for these rows. This does not establish that no alternative exists.
Sources and checks
Read October 22, 2026. Region-scoped rows, not worldwide vendor retirements. Amazon says requests fail on or soon after EOL unless privately arranged. No authenticated endpoint was called. The exact table rows below retain the date, model IDs and Region scope; continuation rows inherit the model named immediately before them. TwelveLabs Marengo Embed 3.0 twelvelabs.marengo-embed-3-0-v1:0 ap-northeast-2 September 22, 2026 March 22, 2027 — | TwelveLabs Pegasus v1.2 twelvelabs.pegasus-1-2-v1:0 af-south-1, ap-east-2, ap-northeast-1, ap-northeast-2, ap-northeast-3, ap-south-1, ap-south-2, ap-southeast-1, ap-southeast-2, ap-southeast-3, ap-southeast-4, ap-southeast-5, ap-southeast-7, ca-central-1, ca-west-1, eu-central-1, eu-central-2, eu-north-1, eu-south-1, eu-south-2, eu-west-1, eu-west-2, eu-west-3, il-central-1, mx-central-1, sa-east-1, us-east-1, us-east-2, us-west-1, us-west-2 September 22, 2026 March 22, 2027 —
Operative page re-read: Read, and it still says what this row says: 28 recorded terms present.
- 145 days
Amazon Web Services · Retirement
The September 30 legacy cohort of third-party Bedrock model IDs reaches scheduled end of life in each row's listed Regions
Check your exact model ID and AWS Region against the operative table. Amazon says migration does not happen automatically; plan a transition to an Active model before the listed EOL date.
Nothing to move to. The captured legacy table has no replacement column and names no successor for these rows. This does not establish that no alternative exists.
Sources and checks
Read October 22, 2026. Region-scoped rows, not worldwide vendor retirements. Amazon says requests fail on or soon after EOL unless privately arranged. No authenticated endpoint was called. The exact table rows below retain the date, model IDs and Region scope; continuation rows inherit the model named immediately before them. DeepSeek DeepSeek-R1 deepseek.r1-v1:0 us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | DeepSeek DeepSeek-V3.1 deepseek.v3-v1:0 (bedrock-runtime), deepseek.v3.1 (bedrock-mantle) ap-northeast-1, ap-south-1, ap-southeast-2, ap-southeast-3, eu-north-1, eu-west-2, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Google Gemma 3 12B IT google.gemma-3-12b-it ap-northeast-1, ap-south-1, ap-southeast-2, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Google Gemma 3 27B PT google.gemma-3-27b-it ap-northeast-1, ap-south-1, ap-southeast-2, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Google Gemma 3 4B IT google.gemma-3-4b-it ap-northeast-1, ap-south-1, ap-southeast-2, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Meta Llama 3 70B Instruct meta.llama3-70b-instruct-v1:0 ap-south-1, ca-central-1, eu-west-2, us-east-1, us-west-2 September 30, 2026 March 30, 2027 To be announced | Meta Llama 3 8B Instruct meta.llama3-8b-instruct-v1:0 ap-south-1, ca-central-1, eu-west-2, us-east-1, us-west-2 September 30, 2026 March 30, 2027 To be announced | Meta Llama 3.1 70B Instruct meta.llama3-1-70b-instruct-v1:0 us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Meta Llama 3.1 8B Instruct meta.llama3-1-8b-instruct-v1:0 us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Meta Llama 3.3 70B Instruct meta.llama3-3-70b-instruct-v1:0 us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Meta Llama 4 Maverick 17B Instruct meta.llama4-maverick-17b-instruct-v1:0 us-east-1, us-east-2, us-west-1, us-west-2 September 30, 2026 March 30, 2027 To be announced | Meta Llama 4 Scout 17B Instruct meta.llama4-scout-17b-instruct-v1:0 us-east-1, us-east-2, us-west-1, us-west-2 September 30, 2026 March 30, 2027 To be announced | MiniMax MiniMax M2 minimax.minimax-m2 ap-northeast-1, ap-south-1, ap-southeast-2, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | MiniMax MiniMax M2.1 minimax.minimax-m2.1 ap-northeast-1, ap-south-1, ap-southeast-2, ap-southeast-3, eu-central-1, eu-north-1, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Mistral AI Devstral 2 123B mistral.devstral-2-123b ap-northeast-1, ap-south-1, ap-southeast-2, ap-southeast-3, eu-central-1, eu-north-1, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Mistral AI Mistral 7B Instruct mistral.mistral-7b-instruct-v0:2 ap-south-1, ap-southeast-2, ca-central-1, eu-west-1, eu-west-2, eu-west-3, sa-east-1, us-east-1, us-west-2 September 30, 2026 March 30, 2027 To be announced | Mistral AI Mixtral 8x7B Instruct mistral.mixtral-8x7b-instruct-v0:1 ap-south-1, ap-southeast-2, ca-central-1, eu-west-1, eu-west-2, eu-west-3, sa-east-1, us-east-1, us-west-2 September 30, 2026 March 30, 2027 To be announced | Mistral AI Voxtral Mini 3B 2507 mistral.voxtral-mini-3b-2507 ap-northeast-1, ap-south-1, ap-southeast-2, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Mistral AI Voxtral Small 24B 2507 mistral.voxtral-small-24b-2507 ap-northeast-1, ap-south-1, ap-southeast-2, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Moonshot AI Kimi K2 Thinking moonshot.kimi-k2-thinking (bedrock-runtime), moonshotai.kimi-k2-thinking (bedrock-mantle) ap-northeast-1, ap-south-1, ap-southeast-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | NVIDIA NVIDIA Nemotron Nano 12B v2 VL BF16 nvidia.nemotron-nano-12b-v2 ap-northeast-1, ap-south-1, ap-southeast-2, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | NVIDIA NVIDIA Nemotron Nano 9B v2 nvidia.nemotron-nano-9b-v2 ap-northeast-1, ap-south-1, ap-southeast-2, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Qwen Qwen3-Coder-30B-A3B-Instruct qwen.qwen3-coder-30b-a3b-v1:0 (bedrock-runtime), qwen.qwen3-coder-30b-a3b-instruct (bedrock-mantle) ap-northeast-1, ap-south-1, ap-southeast-2, ap-southeast-3, eu-central-1, eu-north-1, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Qwen Qwen3 Coder 480B A35B Instruct qwen.qwen3-coder-480b-a35b-v1:0 (bedrock-runtime), qwen.qwen3-coder-480b-a35b-instruct (bedrock-mantle) ap-northeast-1, ap-south-1, ap-southeast-2, ap-southeast-3, eu-north-1, eu-west-2, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Z.AI GLM 4.7 zai.glm-4.7 ap-northeast-1, ap-south-1, ap-southeast-2, ap-southeast-3, eu-north-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced | Z.AI GLM 4.7 Flash zai.glm-4.7-flash ap-northeast-1, ap-south-1, ap-southeast-2, ap-southeast-3, eu-central-1, eu-north-1, eu-south-1, eu-west-1, eu-west-2, sa-east-1, us-east-1, us-east-2, us-west-2 September 30, 2026 March 30, 2027 To be announced
Operative page re-read: Read, and it still says what this row says: 28 recorded terms present.
- 147 days
OpenAI · Retirement
Three models are removed from the API — gpt-5.1, gpt-5.3-codex and gpt-5.4-nano
Search your configuration for the exact ids gpt-5.1, gpt-5.3-codex and gpt-5.4-nano. OpenAI's instruction is to 'Migrate to the recommended replacements before the shutdown date': gpt-6-sol for the first two, gpt-6-luna for gpt-5.4-nano.
Moves to: gpt-6-sol for both gpt-5.1 and gpt-5.3-codex, and gpt-6-luna for gpt-5.4-nano — the ledger's recommended replacement column for all three rows.
Sources and checks
Announced in a ledger section dated October 1, 2026, which states the removal comes 'with six months’ notice'. gpt-5.1 is a floating major id rather than a dated snapshot, so code naming it keeps working to the date with nothing in it that could warn you. Two of the section's three rows — gpt-5.1's and gpt-5.3-codex's — name the same recommended replacement, gpt-6-sol, for a chat model and a code model; the ledger names replacements and does not say they are equivalent to what they replace. The ledger records a schedule and keeps carrying ids after their dates, so it cannot by itself confirm a shutdown happened.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
Passed 32 · 11 filed · 21 not checked
- 3 days ago
xAI · Retirement
The grok-imagine-image-quality model name is retired from the xAI API — and keeps answering, served by a different model
Moves to: grok-imagine-image-2.0 at low quality, which the retired name silently resolves to after the date.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Nothing breaks on the date: 'The slug continues to resolve, so you do not need to change your code to avoid breakage.' What changes is the model behind the name and its price, which xAI states is $0.01 per image lower at every resolution.
Operative page re-read: Read, and it still says what this row says: 3 recorded terms present.
- 5 days ago
OpenAI · Rule due
Existing evals on the deprecated Evals platform become read-only
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
The first of two dated steps for the Evals platform — the shutdown itself is a separate row on November 30. Announced June 3, 2026; the ledger names Promptfoo as a migration path.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 5 days ago
Mistral AI · Retirement
Z.ai GLM 5.2 (zai-glm-5-2), a third-party model served through Mistral's API, is retired
Moves to: Z.ai GLM 5.3 (zai-glm-5-3): 'Use Z.ai GLM 5.3 (zai-glm-5-3) instead, at the same price.' in the changelog, and the Alternative column of the models page's deprecated table.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Mistral's schedule for its own API only; Z.ai's own platform is not covered. Announced September 29, the day after GLM 5.3 became generally available on Mistral, with a month's notice. 'At the same price' is Mistral's statement, not checked against its pricing page, and this desk holds no Mistral key and cannot test whether requests fail.
Operative page re-read: Read, and it still says what this row says: 5 recorded terms present.
- 10 days ago
OpenAI · Rule due
Affected ChatGPT Enterprise workspaces can no longer create new custom GPTs
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Planned, and the FAQ says the Enterprise dates 'are subject to change'. The Internet Archive's September 18 copy of the same FAQ gave September 25 for this cutoff; no release note records the change. Edits to existing, unmigrated GPTs are stated to continue after the cutoff.
Operative page re-read: Read, and it still says what this row says: 10 recorded terms present.
- 13 days ago
OpenAI · Retirement
A second legacy wave shuts down seventeen listed ids: the gpt-3.5/GPT-4-era snapshots (gpt-3.5-turbo-0125, gpt-4-0613, gpt-4-1106-preview, gpt-4-turbo, gpt-4.1-nano, gpt-4o-2024-05-13), the reasoning models o1, o1-pro, o3-mini and o4-mini, the original gpt-image-1 — and six fine-tuned lines including ft-gpt-4, ft-babbage-002 and ft-davinci-002
Moves to: Per id — gpt-5.6-terra for the 3.5 and o4-mini lines, gpt-5.6-sol for the GPT-4 and o1/o3 lines, gpt-5.6-luna for gpt-4.1-nano, gpt-image-2 for gpt-image-1.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
A separate group from the September wave, announced on its own date. The ledger lists each snapshot alias alongside the model name, so a pin on either form is covered. Until August 26 this row described only six of the seventeen ids; the fuller list above is what the cited section has carried since at least August 10.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 14 days ago
Google · Retirement
Three Veo 3.1 preview video models — veo-3.1-generate-preview, veo-3.1-fast-generate-preview and veo-3.1-lite-generate-preview — reach their earliest listed shutdown date in the Gemini API
Moves to: gemini-omni-1.1-flash, named in the deprecations table's replacement column for all three — Google's Gemini Omni Flash video model, not a newer Veo.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Google's lifecycle table says listed shutdown dates 'indicate the earliest possible dates on which a model might be retired'. October 21 update: the pricing page separately states that all three Veo 3.1 previews 'will be shut down on October 22, 2026' and names Gemini Omni Flash or GA models through the Gemini Enterprise Agent Platform as migration options. Read both current pages; no endpoint was called and neither supplies a shutdown hour. Historical reading: the date appeared in the revision stamped 'Last updated 2026-10-01 UTC'; the Internet Archive's September 26 copy said 'No shutdown date announced', and no changelog entry for them was found September 30. Gemini API only; this is not a new reading of Veo availability on Google Cloud's Agent Platform.
Operative page re-read: Read, and it still says what this row says: 24 recorded terms present.
- 14 days ago
Google · Retirement
Gemini Omni Flash Preview reaches its listed Gemini API shutdown date
Moves to: gemini-omni-1.1-flash, named in the lifecycle table
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Read October 22, 2026: the lifecycle table prints gemini-omni-flash-preview June 30, 2026 October 22, 2026 gemini-omni-1.1-flash, under its earliest-possible-date caveat. The pricing page no longer carries this preview ID or yesterday's explicit October 22 warning. Removal from a price page does not prove endpoint shutdown. No endpoint was called, no execution outcome is filed, and the earlier September 30 deprecation remains a distinct historical event.
Operative page re-read: Read, and it still says what this row says: 24 recorded terms present.
- 14 days ago
GitHub · Rule due
Copilot Business and Enterprise: a new default policy starts deciding every eligible generally available feature an administrator has left Unconfigured — and the default is Enabled
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
About feature access, not price. GitHub states explicit decisions are preserved and preview features stay opt-in; the policy also covers the Copilot Code Review policy and the MCP servers in Copilot policy. Which features count as eligible is set by GitHub's docs and can change.
Operative page changed: Read, and something changed: 3 recorded terms no longer on the page. Open the page before relying on this row.
- 15 days ago
Google · Retirement
Sixteen open-model MaaS endpoints retire from Google Cloud's Gemini Enterprise Agent Platform, among them deepseek-ocr-maas, glm-5-maas, gpt-oss-20b-maas, kimi-k2-thinking-maas, llama-3.3-70b-instruct-maas and the qwen3 and multilingual-e5 endpoints
Moves to: Per endpoint, and different for each, the deprecations page's 'Self-deploy alternative': the same model family deployed from Model Garden into your own project — DeepSeek-OCR, DeepSeek-R1, DeepSeek-V3.2, DeepSeek-V3.1, GLM 5, GLM 4.7, GPT OSS, Kimi K2, Llama 3.3, MiniMax M2, E5 Text Embedding (both multilingual-e5 endpoints), Qwen3, Qwen3 Coder and Qwen3 Next (both qwen3-next endpoints). Not a serverless endpoint; Google names no managed replacement, only 'alternative managed endpoints'.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Google's Cloud schedule for its managed endpoints only; the open models stay available to self-deploy, which Google's overview describes as running in your own Google Cloud project, unlike the serverless MaaS endpoints. Unlike the model-versions page, the open-model deprecations page states no rule on whether its dates can move. The alternative column names a model family, not always the exact variant: 'GPT OSS' for gpt-oss-20b-maas, 'Kimi K2' for kimi-k2-thinking-maas. Since the July 21 deprecation, Google says new use of these endpoints 'may be restricted.'
Operative page re-read: Read, and it still says what this row says: 17 recorded terms present.
- 16 days ago
Google · Retirement
gemini-2.5-pro, gemini-2.5-flash and gemini-2.5-flash-lite retire on Google Cloud's Gemini Enterprise Agent Platform (formerly Vertex AI)
Moves to: Per model, from the page's replacement column: gemini-3.5-flash for gemini-2.5-pro; gemini-3.5-flash-lite or gemini-3.1-flash-lite for gemini-2.5-flash; gemini-3.1-flash-lite or Gemma 4 for gemini-2.5-flash-lite.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Google's Cloud schedule only. The page says 'While retirement timelines may be extended, they won't be moved to an earlier date than what is listed', so October 20 is the earliest these can go. The Gemini API deprecations page lists the same three ids with 'No shutdown date announced'.
Operative page re-read: Read, and it still says what this row says: 17 recorded terms present.
- 17 days ago
GitHub · Retirement
GitHub Copilot deprecates GPT-5.5, GPT-5.4, GPT-5.4 mini, GPT-5 mini, Gemini 3.7 Flash and Grok 4.5 across all Copilot experiences
Moves to: GitHub's suggested alternatives: Gemini 3.8 Flash for Gemini 3.7 Flash, GPT-5.6 Sol for GPT-5.5 and GPT-5.4, GPT-5.6 Luna for GPT-5.4 mini and GPT-5 mini, Grok 4.6 for Grok 4.5.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
A retirement from one tool, not from the models' makers: OpenAI's own GPT-5.5 retirement (October 14, from ChatGPT and Codex) 'does not apply to the OpenAI API', and OpenAI names GPT-6 Sol there, not GPT-5.6 Sol. GitHub's supported-models page, read the same evening, carries the October 2 retirements and no 2026-10-19 entry.
Operative page re-read: Read, and it still says what this row says: 7 recorded terms present.
- 21 days ago
Microsoft · Retirement
Azure OpenAI retires sora-2 (version 2025-12-08) — three weeks after OpenAI shut down its own Sora 2 API
Nothing to move to. Microsoft's retirement schedule shows '—' in the Replacement column for sora-2 version 2025-12-08, and its lifecycle page says that when no replacement exists 'customers get 30 days notice before the model retires and inference returns 410 Gone.' OpenAI names no replacement on its side either.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
A Preview-stage 'not sooner than' date, not a committed shutdown. The schedule lists this sora-2 version as Preview, and Microsoft's lifecycle page says Preview models 'launch with a "not sooner than" retirement date (typically 90 days out), but are sometimes extended beyond that initial window, until a suitable replacement preview or GA model version is available.' The same page answers 'Can I get an exception to extend a model's retirement date?' with 'No.' — that is about a customer asking, not about Microsoft moving its own Preview date. Microsoft's schedule (page dated September 21, updated September 23), not an observation: this desk cannot test Azure deployments. The same table lists an earlier sora-2 version, 2025-10-06, with a July 15 retirement date and this version as its replacement.
Operative page changed: Read, and something changed: 1 recorded term no longer on the page. Open the page before relying on this row.
- 31 days ago
OpenAI · Price change
Billing begins for GPT-Rosalind (gpt-rosalind-research), the trusted-access life-sciences model
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Only approved organisations in the trusted-access program can use the model, so almost no reader is affected. The changelog states the rate and the start date; it does not say how usage before October 5 is treated beyond the fact that billing begins then.
Not on the watch list: This page is not on the term-watch list, so nothing here re-reads it on a schedule.
- 31 days ago
Google · Retirement
The managed agent antigravity-preview-05-2026 reaches its listed shutdown date in the Gemini API
Moves to: antigravity-preview-09-2026 — the model page says 'Migrate to antigravity-preview-09-2026', and the deprecations table names it as the recommended replacement.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Google's table says listed shutdown dates 'indicate the earliest possible dates on which a model might be retired', so October 5 is a floor, not a promise of the hour. The changelog, the deprecations table and the agent's own model page all carried the same date and replacement on September 28.
Operative page re-read: Read, and it still says what this row says: 24 recorded terms present.
- 33 days ago
Microsoft · Retirement
Moonshot AI's Kimi-K2.7-Code, a Preview model Azure sells, retires from Microsoft Foundry — the day after GitHub Copilot drops the same model
Nothing to move to. Microsoft's retirement schedule shows '—' in the Replacement column for Kimi-K2.7-Code. GitHub's Copilot notice suggests Kimi K3 for the same model on October 2, but that is GitHub's recommendation for its own tool, not Microsoft's for Foundry.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
A Preview-stage 'not sooner than' date that Microsoft's lifecycle page says is 'sometimes extended beyond that initial window'. The Copilot removal on October 2 is a separate row; Moonshot AI's own API is not covered here.
Operative page changed: Read, and something changed: 1 recorded term no longer on the page. Open the page before relying on this row.
- 34 days ago
GitHub · Retirement
GitHub Copilot deprecates Claude Opus 4.7, Gemini 3.5 Flash, Gemini 3.6 Flash and Kimi K2.7 Code across all Copilot experiences
Moves to: Claude Opus 5 for Opus 4.7, Gemini 3.8 Flash for both Gemini Flash models, Kimi K3 for Kimi K2.7 Code — GitHub's suggested alternatives.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
A retirement from ONE tool, not from the model's maker: Anthropic's deprecations page lists claude-opus-4-7 as Active with a retirement not sooner than April 16, 2027, and Google lists no date for its Gemini 3.x Flash models. The models remain available elsewhere; they leave Copilot.
Operative page re-read: Read, and it still says what this row says: 3 recorded terms present.
- 35 days ago
OpenAI · Retirement
gpt-5.4-cyber is removed from the API
Moves to: 'The most capable cyber model available to you.' — the ledger's replacement column since September 30. Until then it named gpt-5.6-cyber, which OpenAI's model catalog still lists.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Announced in a ledger section dated September 11, 2026 — twenty days of notice, the shortest on the ledger's pending schedule. The replacement it names was rewritten from gpt-5.6-cyber to 'the most capable cyber model available to you' on or about September 30, with no changelog entry; the date stayed October 1. The ledger records a schedule; it keeps carrying ids after their date, so it cannot by itself confirm a shutdown happened.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 35 days ago
Microsoft · Retirement
Microsoft retires its own image models MAI-Image-2.5, MAI-Image-2.5-Flash and MAI-Image-2.5-Pro from Foundry
Nothing to move to. The retirement schedule shows '—' in the Replacement column for all three, although the same table lists MAI-Image-2.5 and MAI-Image-2.5-Flash as the replacements for MAI-Image-2 and MAI-Image-2e, retired August 15. Microsoft's MAI image how-to page lists the Preview models MAI-Image-2.6 and MAI-Image-2.6-Flash without calling them replacements.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
A Preview-stage 'not sooner than' date: all three are Preview models, and Microsoft's lifecycle page says Preview models 'launch with a "not sooner than" retirement date (typically 90 days out), but are sometimes extended beyond that initial window'. This desk cannot test Foundry deployments or see the 30-day notice the policy promises when no replacement exists.
Operative page changed: Read, and something changed: 1 recorded term no longer on the page. Open the page before relying on this row.
- 36 days ago
Mistral AI · Retirement
The free Labs model Leanstral 1.5 (labs-leanstral-1-5) is retired from Mistral's API
Nothing to move to. Neither announcement names a replacement: not the June 30 release line, and not the September 29 changelog entry that deprecates it, though the OCR 4.0 and Z.ai GLM 5.2 entries beside it each name one. Mistral's models page now lists Leanstral 1.5 in its deprecated table (deprecated 9/29/2026, retired 9/30/2026) with the Alternative column empty. The Labs model before it, labs-leanstral-2603, was retired on June 30 with Leanstral 1.5 as its alternative; nothing points past this one.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
First stated in a release-day changelog line, 'This model will be retired on September 30, 2026.', while the models page listed it as current. On September 29 Mistral confirmed the date: a changelog entry calls it 'an experimental Labs model' that 'is deprecated and retires on September 30, 2026', and the models page's deprecated table gives the same day — one day after the deprecation date it lists. A free preview model; nothing here is a price.
Operative page re-read: Read, and it still says what this row says: 5 recorded terms present.
- 36 days ago
Google · Retirement
The gemini-omni-flash-preview endpoint is deprecated in the Gemini API
Moves to: gemini-omni-1.1-flash, released August 27, 2026 as the GA version of the same model.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Google's sentence says the endpoint 'will be deprecated', not removed, and names the GA model in the neighbouring entry rather than as a stated replacement. The changelog redirects an unauthenticated plain client to a sign-in page, so this desk reads it through a rendering fetcher.
Operative page re-read: Read, and it still says what this row says: 10 recorded terms present.
- 36 days ago
Amazon · Retirement
Amazon Bedrock ends its own Nova Canvas image model and Nova Reel video model (amazon.nova-canvas-v1:0, amazon.nova-reel-v1:0 and amazon.nova-reel-v1:1)
Nothing to move to. Bedrock's legacy lifecycle table has no replacement column, and neither the Nova Canvas nor the Nova Reel model card names a successor; the table's rule for the date is 'Migration will not happen automatically.' A reading of the pages, not of what Amazon may tell customers directly.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
Sources and checks
Amazon's schedule; both models have been Legacy since March 30, 2026, and this desk holds no AWS account and cannot test whether either model answers. The page says 'For Amazon Bedrock usage, only the dates on this page apply.' The same table says it leaves out models past their end-of-life date, yet on September 28 still listed Nova Premier and Nova Sonic, whose date was September 14 — so a row still there after September 30 settles nothing either way.
Operative page re-read: Read, and it still says what this row says: 28 recorded terms present.
- 38 days ago
OpenAI · Retirement
Legacy snapshots gpt-3.5-turbo-instruct, babbage-002, davinci-002 and gpt-3.5-turbo-1106 shut down
Moves to: gpt-5.6-terra, named for all four ids.
Not establishable by this deskFiled as past by OpenAI's own ledger; the shutdown itself is not confirmed here. The four ids sit under 'Past deprecations' as of September 30 with the September 28 date and gpt-5.6-terra unchanged, while the section's paragraph still says access 'will be shut down'. No request was made to the models.
Read the entry that records itSources and checks
Part of a stated multi-wave retirement of older models with declining usage. A substitute model is named for every id in this group.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 42 days ago
OpenAI · Retirement
The Videos API and the entire Sora 2 model family shut down
Nothing to move to. The ledger's replacement column reads "---" for all six rows, and no migration sentence appears anywhere in the announcement. Measured 2026-09-07: these are the only six rows on the page in that condition, out of 49 with a replacement column.
Confirmed by this desk's own re-readShut down, by OpenAI's own account and by wire. The sora-2 model page now reads 'The Sora 2 models and Videos API were shut down on September 24, 2026 and are no longer available. There is no one-to-one replacement API.', and unauthenticated GET and POST /v1/videos answer 404 with an empty body — the answer this API gives a route that does not exist — while /v1/models answers 401. The ledger this row cites still lists all six rows under 'Upcoming deprecations' three days after the date, and the help centre still says the API 'will be discontinued'.
Read the entry that records itSources and checks
Covers the Videos API plus sora-2, sora-2-pro and three dated snapshots. The consumer app went earlier, on April 26, 2026; the help-centre article for it commits to deleting Sora content after a final export window it describes only conditionally.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 53 days ago
Anthropic · Promotional window ends
The 50% Claude Code weekly-limit boost ends
Confirmed by this desk's own re-readEnded at its stated 11:59 PM PT on September 13, when the terms article was rewritten — and limits did not fall back: 'starting September 14, 2026, weekly limits in Claude Code are 25% higher than they were before the promotion for Pro, Max, Team, and seat-based Enterprise plans.'
Read the entry that records itSources and checks
Extended three times after its first stated end (July 13, then August 19, August 31 and September 13); the terms article is the page that moved each time. The stretch between the first window's stated July 13 end and the July 18 renewal has no provider-authored record and is carried at coverage tier only.
Operative page changed: Read, and something changed: 1 recorded term no longer on the page. Open the page before relying on this row. Historical reading carried forward; this sweep did not re-read these terms.
- 66 days ago
OpenAI · Retirement
GPT-5.4 and GPT-5.4 mini stop being available in Codex for users signed in with ChatGPT
Moves to: A recommended replacement model in Codex, or the same models via API-key authentication — the retirement is scoped to an access path, not to the model.
Confirmed by this desk's own re-readThe page still describes it as upcoming. Read 2026-09-14: the Codex models page, since reworded, says 'GPT-5.4 and GPT-5.4 mini retire from Codex on August 31, 2026' and still offers the codex -m gpt-5.4 and codex -m gpt-5.4-mini commands. The page was not updated; whether ChatGPT sign-in access actually ended is not something it settles.
Read the entry that records itSources and checks
Scoped to an access path, not the models: both remain on the OpenAI API and in Codex sessions authenticated with an API key.
Operative page re-read: Read, and it still says what this row says: 4 recorded terms present.
- 66 days ago
Anthropic · Price change
Sonnet 5 introductory API pricing ($2/$10) ends
Confirmed by this desk's own re-readReversed: the introductory $2/$10 rate 'is now the standard price', and the page says the 'previously scheduled increase to $3/$15 per million input/output tokens on September 1, 2026 will not occur.' First found on September 1; unchanged on 2026-09-14.
Read the entry that records itSources and checks
Standard pricing of $3/$15 is stated for September 1. These are the provider's published terms, not a quote for any account.
Operative page re-read: Last recorded read contained 2 recorded terms present. Historical reading carried forward; this sweep did not re-read these terms.
- 67 days ago
OpenAI · Retirement
The official DALL·E GPT is retired in ChatGPT
Moves to: Image generation in ChatGPT itself; the notice states user-created GPTs with image generation enabled are unaffected.
Not establishable by this deskNot establishable. The ChatGPT release notes, readable on 2026-09-14 after refusing this desk's client through September 1, still carry only the notice 'On August 30, 2026, we're retiring the official DALL·E GPT in ChatGPT'; nothing after it records the retirement.
Read the entry that records itSources and checks
User-created GPTs with image generation enabled are stated to be unaffected. ChatGPT Images remains available for creating and editing.
Operative page re-read: Last recorded read contained 2 recorded terms present. Historical reading carried forward; this sweep did not re-read these terms.
- 71 days ago
OpenAI · Retirement
The Assistants API shuts down on two vendors at once — OpenAI names the Responses and Conversations APIs as the replacement, Microsoft retires Azure OpenAI Assistants the same day and names Microsoft Foundry Agents instead
Moves to: The Responses or Conversations API for direct OpenAI integrations; Azure customers are pointed at Microsoft's own path, which is a different vendor's recommendation.
Confirmed by this desk's own re-readShut down, and establishable from outside: unauthenticated, /v1/assistants returns 404 with an empty body — the same answer this API gives for a route that does not exist — while a live route still answers 401 with a JSON body. Six spaced probes across 2026-08-26 and 2026-08-27, each run with the missing-route control alongside, never moved. Microsoft's Foundry (classic) notice, future-tense on the morning of the 26th, now reads "The Assistants API is retired."
Read the entry that records itSources and checks
An API shutdown affects integrations rather than chat users, and both replacements are their vendor's own recommendation rather than this desk's advice. Added 2026-08-19: this row previously carried only the OpenAI exit, which was accurate for OpenAI's surface and silent about Azure. Microsoft's Foundry (classic) Assistants page, read the same day, states the feature "is deprecated and will be retired on August 26, 2026" and directs workloads to Microsoft Foundry Agents — https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/assistants?view=foundry-classic
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 71 days ago
OpenAI · Retirement
OpenAI o3 is retired from ChatGPT after a stated 90-day sunset
Moves to: GPT-5.6 in ChatGPT — the release notes retire o3 into the current default rather than naming a per-model substitute.
Not establishable by this deskNot establishable. Read 2026-09-14, nineteen days after the date: the model release notes still say 'OpenAI o3 will be retired from ChatGPT on August 26, 2026', in the future tense, and no later entry records the retirement. A schedule nobody updated cannot show whether it was kept.
Read the entry that records itSources and checks
A retirement notice records that access is ending. It says nothing about the model's quality or about why.
Operative page re-read: Last recorded read contained 1 recorded term present. Historical reading carried forward; this sweep did not re-read these terms.
- 78 days ago
Anthropic · Promotional window ends
Doubled Cowork 5-hour usage limits end and standard limits resume
Not establishable by this deskNot establishable. The terms article that carried the window has answered HTTP 404 since August 20 and still did on 2026-09-14; no replacement terms article was found, and the nearest help article describes how Cowork uses limits without stating them.
Read the entry that records itSources and checks
A promotional window ending is a return to standard terms, not a reduction of anything anyone was entitled to. The launch post still carries the superseded August 5 date; the terms article is the one that gets edited.
Cannot be checked from here: Not checkable from here. neither client read it on any of 3 spaced attempts — first client: HTTP 404, HTTP 404, HTTP 404; second client: curl answered HTTP 404, curl answered HTTP 404, curl answered HTTP 404
- 87 days ago
OpenAI · Retirement
gpt-5.2-chat-latest and gpt-5.3-chat-latest shut down on the API, replaced by gpt-5.6-sol
Moves to: gpt-5.6-sol, named in the ledger's replacement column for both aliases.
Confirmed by this desk's own re-readExecuted on schedule, 94 days after the May 8 notice. The ledger still carries both ids against "Aug 10, 2026" with gpt-5.6-sol as the replacement; neither appears on the live models list, and no "-chat-latest" alias does.
Read the entry that records itSources and checks
Scoped to the two dated chat-latest aliases. The deprecations ledger is the provider's own schedule and is where any change to this date would appear.
Operative page re-read: Read, and it still says what this row says: 29 recorded terms present.
- 88 days ago
OpenAI · Retirement
The Atlas browser stops working
Moves to: ChatGPT for browser-based agentic work, named in the article's own title.
Not establishable by this deskReported to have happened, but not establishable from any OpenAI page this desk can read: help.openai.com, openai.com and chatgpt.com all refuse or challenge the identifying client, while the developer docs serve it the whole document.
Read the entry that records itSources and checks
The release notes say Atlas is scheduled to stop working on this date. A scheduled shutdown is a stated plan, not a guarantee of the hour.
Operative page re-read: Last recorded read contained 1 recorded term present. Historical reading carried forward; this sweep did not re-read these terms.
Government obligations
A second board, kept separate on purpose. These are duties a government body has published in its own document — a comment period closing, something an agency owes by a date. The board above is providers only, and mixing the two would make a list that means one thing mean two. 6 of these 9 dates are not printed anywhere: the order states a clock (“within 90 days of the date of this order”) and the date is this desk’s arithmetic, shown on every row and re-derived by the test suite.
- 23 days
Agriculture and HHS · Owed by an agency
Explore options for labeling food and drug products containing significant amounts of alpha-galactose
Counted by this desk, and it lands on a Saturday. The order says "within 60 days of the date of this order" and sets no business-day rule, so nothing here says the deadline moves.
The document
Within 60 days of the date of this order, the Secretary of Agriculture and the Secretary of Health and Human Services, as appropriate, shall explore options for labeling food and drug products that include significant amounts of alpha-galactose, in a manner similar to other allergy warnings
Within 60 days of the date of this order, signed Sep 29, 2026 — this desk counts Nov 28, 2026.
The order requires exploring options, not labeling anything, and qualifies even that with 'as appropriate'. Nothing here obliges either department to publish what it finds.
- 23 days
Environmental Protection Agency, in consultation with Agriculture and HHS · Owed by an agency
Submit to the President, through the Homeland Security Advisor, a regulatory framework for releasing advanced tick and mosquito control technologies into the environment
Counted by this desk, and it lands on a Saturday. The order says "within 60 days of the date of this order" and sets no business-day rule, so nothing here says the deadline moves.
The document
Within 60 days of the date of this order, to expedite the campaign ordered in subsection 3(a) of this order, the Administrator of the Environmental Protection Agency, in consultation with the Secretary of Agriculture and the Secretary of Health and Human Services, shall: (i) submit to the President, through the Assistant to the President and Homeland Security Advisor, a regulatory framework, informed by relevant academic centers and industry, for the planned development and release into the environment of advanced, safe, and effective tick and mosquito control technologies
Within 60 days of the date of this order, signed Sep 29, 2026 — this desk counts Nov 28, 2026.
A framework submitted to the President is not a published rule, and the order does not require it to be released. The same subsection also tells EPA to advance or register novel control methods it finds safe and effective, which carries no date and is not tracked here.
- 23 days
Interior, Agriculture, EPA and GSA, in consultation with HHS · Owed by an agency
Develop and implement the DC-wide tick and invasive-mosquito campaign, including a plan for controls on lands across DC communities
Counted by this desk, and it lands on a Saturday. The order says "within 60 days of the date of this order" and sets no business-day rule, so nothing here says the deadline moves.
The document
Within 60 days of the date of this order, the Secretary of the Interior, the Secretary of Agriculture, the Administrator of the Environmental Protection Agency, and the Administrator of General Services, in consultation with the Secretary of Health and Human Services, shall develop and implement a campaign across DC
Within 60 days of the date of this order, signed Sep 29, 2026 — this desk counts Nov 28, 2026.
The 60 days run from the September 29 signing date, not from the October 2 printing, and the order sets no business-day rule — this desk's count lands on a Saturday. The campaign's own targets (90 percent of invasive mosquitoes, 50 percent of ticks) are due by 2028, not by this date; what is due here is developing and implementing the campaign.
- 26 days
Federal Reserve Board · Comments close
Comments close on the model changes proposed for the Board's 2027 supervisory stress test
Comment by December 1 under Docket No. OP-1882. The notice's preferred method is the Board's own proposals form at federalreserve.gov/apps/proposals.
The document
Comments must be received on or before December 1, 2026.
The document prints the date: Dec 1, 2026.
A comment period closing, not a rule taking effect. The notice is a request for comment on model changes; nothing in it binds anyone on December 1 except the Board's own willingness to read late comments.
- 53 days
Agriculture, with Interior, HHS and EPA · Owed by an agency
Issue prize competition announcements or direct research programs for non-chemical means of reducing tick- and mosquito-borne diseases
The document
Within 90 days of the date of this order, the Secretary of Agriculture, in coordination with the Secretary of the Interior, the Secretary of Health and Human Services, and the Administrator of the Environmental Protection Agency, shall issue prize competition announcements under 15 U.S.C. 3719 or other applicable laws, or direct research programs, for the development of non-chemical means
Within 90 days of the date of this order, signed Sep 29, 2026 — this desk counts Dec 28, 2026.
Shares its deadline and its statutory authority with the HHS row above but not its subject: this one is non-chemical population control, that one is diagnostics and treatment. Either could be satisfied by a directed program that is never announced.
- 53 days
Health and Human Services · Owed by an agency
Issue prize competition announcements or direct R&D programs for diagnostics, prophylactics and therapeutics for tick- and mosquito-borne diseases, and review alpha-gal immunotherapy candidates
The document
Within 90 days of the date of this order, the Secretary of Health and Human Services, building on the Lyme Disease Moonshot initiative already underway, as appropriate, shall: (i) issue prize competition announcements under 15 U.S.C. 3719 or other applicable laws, or direct research and development programs
Within 90 days of the date of this order, signed Sep 29, 2026 — this desk counts Dec 28, 2026.
The order offers alternatives — prize announcements under 15 U.S.C. 3719, other applicable laws, or directed programs — so several different acts would satisfy it, and only the first is the kind that is announced publicly. Which diseases count is left to the head of the agency.
- 53 days
Office of Management and Budget · Owed by an agency
The Director of OMB issues the memorandum to agency heads on implementing the America.gov order
The document
Within 90 days of the date of this order, the Director of OMB shall issue a memorandum to the heads of agencies on implementation of this order.
Within 90 days of the date of this order, signed Sep 29, 2026 — this desk counts Dec 28, 2026.
The order does not require the memorandum to be published, and OMB memoranda are not always posted promptly. It is also the document that would say which services are covered and which are excluded, so the board watching this date is watching the thing that makes the rest of the order legible.
Passed 2 · 0 filed · 2 not checked
- Yesterday
Federal Reserve Board · Comments close
Comments close on the proposal to modernize Regulation O, the rule governing credit a bank extends to its own executives, directors and principal shareholders
Moved from Oct 5, 2026. Federal Reserve Board announced the extension on Oct 2, 2026. The new date is not in the Federal Register yet — it comes from the announcement quoted below.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
The document
Comments must be submitted on or before October 5, 2026.
The document prints the date: Oct 5, 2026.
The Federal Reserve Board announced Friday that it will extend, until November 4, the comment period on its proposal to modernize Regulation O, its rule governing the extension of credit by a bank to its "insiders"—bank executives, board members and major shareholders who could potentially influence a bank's lending decisions.
Federal Reserve Board, Oct 2, 2026 — the board counts Nov 4, 2026.
The printed deadline is October 5; the November 4 date comes from the Board's own press release of October 2, which says it WILL extend. The release does not give the year, and nothing this desk read states a new date in the Federal Register — so a commenter relying on November 4 is relying on an announcement, not on a notice. The Board also says it extended the period to give parties more time to prepare comments; nothing in either document changes what the proposal would do.
- 3 days ago
Census Bureau · Comments close
Comments close on the proposed residence criteria for the next decennial census and the proposed regulations for its demographic questions
Moved from Oct 13, 2026. Census Bureau announced the extension on Oct 2, 2026. The new date is not in the Federal Register yet — it comes from the announcement quoted below.
This date passed and nothing is filedThe board is not claiming it happened. This desk has not checked.
The document
To ensure consideration during the decision-making process, comments must be received by October 13, 2026.
The document prints the date: Oct 13, 2026.
The proposed rule's comment period, which would have ended on October 13, 2026, is extended until November 2, 2026.
Census Bureau, Oct 2, 2026 — the board counts Nov 2, 2026.
Unlike the Regulation O row, the extension here is a Federal Register document rather than a press release — it just has not printed yet. FR Doc. 2026-20328 was filed for public inspection on October 2 and is scheduled to publish on October 5 at 91 FR 63207, and its own DATES line reads 'Submit comments on or before November 2, 2026.' Re-read October 4, the proposed rule's own Register record still returns a comment close date of October 13 and an empty corrections list. A docket extension is only more time to comment; nothing in either document changes what the rule would do, and neither says the Bureau will extend again.