KINGY / AI DEADLINESSource records ↗

DON’T LET A RETIREMENT BREAK YOUR WORKFLOW

Your AI stack.
Your next deadline.

Know what’s changing, which route it affects and when to act. Keep a calendar or get the reminders you choose.

Kingy tracks 399 AI model retirements and platform deadlines across 10 providers and clouds, including OpenAI, Anthropic, Google, xAI, Mistral and DeepSeek. Each links to the official notice. As of November 22, 2026, 178 have a date still ahead; the next is November 24, 2026. How Kingy checks AI facts.

Loading the source catalog…

Take the dates with you

All-day planning markers for Apple Calendar, Google Calendar and Outlook. No invented shutdown time.

Next 40 AI deadlines as of November 22, 2026

Showing 40 of 178 upcoming deadlines. Filters, watchlists, calendars and the full catalog load with the tracker. Dates are planning markers; the vendor notice controls.

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2026-11-24Listed date

claude-opus-4-5 (version 1)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule, Anthropic-hosted version 1

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule, Anthropic-hosted version 1; model version 1. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Amazon Bedrock · Scheduled · SOURCE REVIEW REQUIRED
2026-11-26Listed date

Jamba 1.5 Large

Route: Amazon Bedrock, regions: us-east-1

Amazon Bedrock, regions: us-east-1. Other AWS regions and provider-direct APIs can have different dates.

AWS says that on or soon after EOL, the version becomes unavailable in the covered regions and requests fail unless a private continuation arrangement applies. Migration is not automatic. No runtime test.

Amazon Bedrock · Scheduled · SOURCE REVIEW REQUIRED
2026-11-26Listed date

Jamba 1.5 Mini

Route: Amazon Bedrock, regions: us-east-1

Amazon Bedrock, regions: us-east-1. Other AWS regions and provider-direct APIs can have different dates.

AWS says that on or soon after EOL, the version becomes unavailable in the covered regions and requests fail unless a private continuation arrangement applies. Migration is not automatic. No runtime test.

Amazon Bedrock · Scheduled · SOURCE REVIEW REQUIRED
2026-11-27Listed date

Claude 3 Haiku

Route: Amazon Bedrock, regions: ca-central-1

Amazon Bedrock, regions: ca-central-1. Other AWS regions and provider-direct APIs can have different dates.

AWS says that on or soon after EOL, the version becomes unavailable in the covered regions and requests fail unless a private continuation arrangement applies. Migration is not automatic. No runtime test.

Amazon Bedrock · Scheduled · SOURCE REVIEW REQUIRED
2026-11-27Listed date

Claude 3 Sonnet

Route: Amazon Bedrock, regions: ca-central-1

Amazon Bedrock, regions: ca-central-1. Other AWS regions and provider-direct APIs can have different dates.

AWS says that on or soon after EOL, the version becomes unavailable in the covered regions and requests fail unless a private continuation arrangement applies. Migration is not automatic. No runtime test.

OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2026-11-30Listed date

OpenAI Agent Builder shutdown deadline: November 30, 2026

Agent Builder

Export and inventory the workflow, choose the Agents SDK or a suitable Workspace Agent, then validate tools, state, permissions, guardrails and rollback before the November 30 shutdown.

Migration guidance: Agents SDK or ChatGPT Workspace Agents

Anthropic · Tentative
2026-11-30Listed date

Claude Sonnet 4.5

Route: Claude API (Anthropic direct)

claude-sonnet-4-5-20250929 on the Claude API (Anthropic direct). Partner-operated platforms have separate schedules.

Anthropic lists this model as deprecated and gives a tentative retirement date. Requests to retired Claude API models fail; no runtime test performed.

Migration guidance: claude-sonnet-5-5

OpenAI · Scheduled
2026-11-30Listed date

OpenAI Evals shutdown

Evals dashboard and API; read-only phase begins October 31.

Test the replacement with your own workload and migrate before the listed date.

Migration guidance: Migrate evaluation workflows; vendor guide links Promptfoo.

OpenAI · Scheduled
2026-11-30Listed date

OpenAI reusable prompt objects

v1/prompts API and reusable prompt objects

Test the replacement with your own workload and migrate before the listed date.

Migration guidance: Move prompt content into application code.

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2026-11-30Listed date

claude-sonnet-4-5 (version 1)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule, Anthropic-hosted version 1

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule, Anthropic-hosted version 1; model version 1. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Migration guidance: claude-sonnet-5-5

Amazon Bedrock · Scheduled · SOURCE REVIEW REQUIRED
2026-11-30Listed date

Marengo Embed v2.7

Route: Amazon Bedrock, regions: ap-northeast-2, eu-west-1, us-east-1

Amazon Bedrock, regions: ap-northeast-2, eu-west-1, us-east-1. Other AWS regions and provider-direct APIs can have different dates.

AWS says that on or soon after EOL, the version becomes unavailable in the covered regions and requests fail unless a private continuation arrangement applies. Migration is not automatic. No runtime test.

OpenAI · Scheduled
2026-12-01Listed date

Three GPT Image API models are scheduled to retire December 1, 2026

gpt-image-1-mini, gpt-image-1.5, chatgpt-image-latest

Inventory the three retiring IDs and test the currently named Sunburst or Flare replacement before December 1. Retain the earlier GPT Image 2 guidance in correction history; do not transfer version-specific settings.

Migration guidance: gpt-image-2.5-sunburst or gpt-image-2.5-flare; validate the exact workflow.

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2026-12-03Listed date

DeepSeek-V4-Flash-0731 (version 2026-07-31)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version 2026-07-31. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2026-12-08Listed date

gpt-chat-latest (version 2026-05-28)

Route: Azure OpenAI, public cloud; deployment type not specified by the retirement schedule

Azure OpenAI, public cloud; deployment type not specified by the retirement schedule; model version 2026-05-28. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists lifecycle state Preview. Its lifecycle policy says retired models are removed from service and inference returns HTTP 410 Gone. No runtime availability test; do not assume a passed date confirms current service state.

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2026-12-09Listed date

Ministral-3B (version 1)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version 1. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Migration guidance: ministral-ministral-4b-instruct

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2026-12-09Listed date

mistral-medium-2505 (version 1)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version 1. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Migration guidance: mistral-medium-3-5

OpenAI · Scheduled
2026-12-11Listed date

GPT-5 mini

Route: OpenAI direct API

OpenAI direct API: gpt-5-mini and gpt-5-mini-2025-08-07

OpenAI lists this API model for shutdown on December 11, 2026. No runtime test performed.

Migration guidance: gpt-5.6-terra

Google Vertex AI · Scheduled · SOURCE REVIEW REQUIRED
2026-12-13Listed date

gemini-live-2.5-flash-native-audio

Route: Google Vertex AI / Gemini Enterprise Agent Platform, Google-managed model API

Google-managed cloud model API: gemini-live-2.5-flash-native-audio. The Gemini developer API and partner-operated routes have separate schedules.

Google lists this cloud-route retirement marker. Timelines may be extended; an elapsed date alone does not prove runtime unavailability.

Migration guidance: gemini-3.8-live

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2026-12-15Listed date

tts (version 001)

Route: Azure OpenAI, public cloud; deployment type not specified by the retirement schedule

Azure OpenAI, public cloud; deployment type not specified by the retirement schedule; model version 001. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists lifecycle state Preview. Its lifecycle policy says retired models are removed from service and inference returns HTTP 410 Gone. No runtime availability test; do not assume a passed date confirms current service state.

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2026-12-15Listed date

tts-hd (version 001)

Route: Azure OpenAI, public cloud; deployment type not specified by the retirement schedule

Azure OpenAI, public cloud; deployment type not specified by the retirement schedule; model version 001. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists lifecycle state GA. Its lifecycle policy says retired models are removed from service and inference returns HTTP 410 Gone. No runtime availability test; do not assume a passed date confirms current service state.

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2026-12-15Listed date

whisper (version 001)

Route: Azure OpenAI, public cloud; deployment type not specified by the retirement schedule

Azure OpenAI, public cloud; deployment type not specified by the retirement schedule; model version 001. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists lifecycle state GA. Its lifecycle policy says retired models are removed from service and inference returns HTTP 410 Gone. No runtime availability test; do not assume a passed date confirms current service state.

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2026-12-15Listed date

Cohere-parse-v5 (version 1)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version 1. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2026-12-16Listed date

gpt-image-1.5 (version 2025-12-16)

Route: Azure OpenAI, public cloud; deployment type not specified by the retirement schedule

Azure OpenAI, public cloud; deployment type not specified by the retirement schedule; model version 2025-12-16. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists lifecycle state GA. Its lifecycle policy says retired models are removed from service and inference returns HTTP 410 Gone. No runtime availability test; do not assume a passed date confirms current service state.

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2027-01-05Listed date

FLUX-1.1-pro (version 1)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version 1. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2027-01-05Listed date

FLUX.1-Kontext-pro (version 1)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version 1. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

OpenAI · Scheduled
2027-01-06Listed date

OpenAI text-to-speech models

API: tts-1, tts-1-hd, gpt-4o-mini-tts-2025-03-20, gpt-4o-mini-tts-2025-12-15

Test the replacement with your own workload and migrate before the listed date.

Migration guidance: gpt-realtime-2.1-mini

Amazon Bedrock · Scheduled · SOURCE REVIEW REQUIRED
2027-01-08Listed date

Claude Opus 4.1

Route: Amazon Bedrock, regions: us-east-1, us-east-2, us-west-2

Amazon Bedrock, regions: us-east-1, us-east-2, us-west-2. Other AWS regions and provider-direct APIs can have different dates.

AWS says that on or soon after EOL, the version becomes unavailable in the covered regions and requests fail unless a private continuation arrangement applies. Migration is not automatic. No runtime test.

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2027-01-12Listed date

gpt-chat-latest (version 2026-08-06)

Route: Azure OpenAI, public cloud; deployment type not specified by the retirement schedule

Azure OpenAI, public cloud; deployment type not specified by the retirement schedule; model version 2026-08-06. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists lifecycle state Preview. Its lifecycle policy says retired models are removed from service and inference returns HTTP 410 Gone. No runtime availability test; do not assume a passed date confirms current service state.

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2027-01-14Listed date

gpt-4o (version 2024-05-13)

Route: Azure OpenAI, Global Standard

Azure OpenAI, Global Standard; model version 2024-05-13. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft explicitly says this deployment type automatically upgrades to gpt-5.1 version 2025-11-13 at retirement.

Migration guidance: gpt-5.1 (2025-11-13)

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2027-01-14Listed date

gpt-4o (version 2024-05-13)

Route: Azure OpenAI, Data Zone Standard

Azure OpenAI, Data Zone Standard; model version 2024-05-13. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft explicitly says this deployment type automatically upgrades to gpt-5.1 version 2025-11-13 at retirement.

Migration guidance: gpt-5.1 (2025-11-13)

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2027-01-14Listed date

gpt-4o (version 2024-05-13)

Route: Azure OpenAI, Regional Provisioned

Azure OpenAI, Regional Provisioned; model version 2024-05-13. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft says provisioned deployments are not automatically upgraded. Create replacement deployments and move traffic before retirement.

Migration guidance: gpt-5.1 (2025-11-13)

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2027-01-14Listed date

gpt-4o (version 2024-05-13)

Route: Azure OpenAI, Global Provisioned

Azure OpenAI, Global Provisioned; model version 2024-05-13. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft says provisioned deployments are not automatically upgraded. Create replacement deployments and move traffic before retirement.

Migration guidance: gpt-5.1 (2025-11-13)

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2027-01-14Listed date

gpt-4o (version 2024-05-13)

Route: Azure OpenAI, Data Zone Provisioned

Azure OpenAI, Data Zone Provisioned; model version 2024-05-13. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft says provisioned deployments are not automatically upgraded. Create replacement deployments and move traffic before retirement.

Migration guidance: gpt-5.1 (2025-11-13)

Azure OpenAI · Scheduled · SOURCE REVIEW REQUIRED
2027-01-14Listed date

gpt-4o (version 2024-05-13)

Route: Azure OpenAI, Global Batch

Azure OpenAI, Global Batch; model version 2024-05-13. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists lifecycle state Deprecated. Its lifecycle policy says retired models are removed from service and inference returns HTTP 410 Gone. No runtime availability test; do not assume a passed date confirms current service state.

Migration guidance: gpt-5.1 (2025-11-13)

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2027-01-15Listed date

claude-opus-5-5-fast (version 2)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule, Azure-hosted version 2

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule, Azure-hosted version 2; model version 2. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2027-01-17Listed date

Phi-4 (version —)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version —. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Migration guidance: microsoft--phi-4 (Foundry Managed Compute)

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2027-01-17Listed date

Phi-4-mini-instruct (version —)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version —. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Migration guidance: microsoft--phi-4 (Foundry Managed Compute)

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2027-01-17Listed date

Phi-4-mini-reasoning (version —)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version —. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Migration guidance: microsoft--phi-4 (Foundry Managed Compute)

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2027-01-17Listed date

Phi-4-multimodal-instruct (version —)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version —. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Migration guidance: microsoft--phi-4 (Foundry Managed Compute)

Azure AI Foundry · Scheduled · SOURCE REVIEW REQUIRED
2027-01-17Listed date

Phi-4-reasoning (version —)

Route: Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule

Azure AI Foundry, public cloud; deployment type not specified by the retirement schedule; model version —. Do not apply to provider-direct APIs, government clouds or fine-tuned models.

Microsoft lists this model and version in its Foundry lifecycle schedule. Follow the linked official schedule and deployment-specific guidance. This record does not assert a runtime error code or automatic migration for this route.

Migration guidance: microsoft--phi-4 (Foundry Managed Compute)

ONLY IF YOU ASK

Email alerts for your stack

Choose providers and reminder windows. Confirm your email before any alerts start. Optional notice-change alerts cover the selected catalog. Unsubscribe from every alert.

Your watchlist stays in your browser. Email and preferences are stored privately on Kingy’s hosting, separately from newsletters. Unconfirmed requests expire after 48 hours; unsubscribe deletes your record. Mail acceptance does not guarantee inbox delivery.

Providers
Days before the listed date

Sources, date precision and how alerts work

This selected catalog combines retained Stack Change Radar records with official notices reviewed for this release. It is not a census of all providers. Read the route and source on each record: an API retirement may not affect the consumer app or another cloud provider.

Read each record’s date type and qualifier. “Not sooner than” is a minimum lifetime, not a scheduled retirement. An elapsed date alone does not prove a service is offline. Unknown dates have no calendar event. Records needing source review are excluded from calendars and email reminders.

Calendar downloads are snapshots. Subscribe to the copied calendar URL for updates; your calendar controls polling. Stable event IDs and revision numbers allow updates. A removed date becomes a cancelled calendar event. Reminder support varies by calendar.

Email reminders run hourly through the existing hosting scheduler. They are sent on the selected day at 30, 7 or 1 days before a listed date, at most once per record revision and window. There is no backfill of missed reminder days. Existing automated source wording checks run daily. The newly added provider and cloud sources require manual re-review after seven days; their checks do not crawl those sites automatically. Changed or stale evidence is held for review. Notice-change alerts follow catalog updates, not every edit to a vendor page. No AI inference testing is claimed.

Browse the lifecycle archive · Report a correction · Kingy privacy policy

Cloud lifecycle summaries credit Amazon Web Services (CC BY-SA 4.0, summarized with route/date fields), Google (CC BY 4.0, summarized), and Microsoft Learn. The original source controls. The Bedrock-derived summaries and factual review receipts are available under CC BY-SA 4.0.

Why counts differ

Radar counts individual model and route lifecycle facts. AI Deadlines is a selected catalog of actionable notices, sometimes grouping several model IDs in one notice. The open shutdowns dataset also covers products and plans. These counts measure different things. The former Deadlines import refreshed existing records but omitted new Radar records; additions are now reviewed against official sources.