November 2026 AI model and developer platform schedule: Sonnet 4.5 retirement, GPT-5.6 Sol promo pricing, OpenAI shutdowns and three conferences
As of 2026-10-26, three more November dates are officially announced: DeepSeek Search/Fetch start billing per successful request at 00:00 Beijing time on 2026-11-01; the full Kimi K3.1 Pro weights will be released by 2026-11-04; gemini-3.1-flash-tts-preview shuts down on 2026-11-17. The earlier dates still hold: Claude Sonnet 4.5 (claude-sonnet-4-5-20250929) retires from the Claude API on 2026-11-30, with claude-sonnet-5-5 as the recommended replacement; GPT-5.6 Sol's promotional pricing is available “until” 2026-11-30; OpenAI shuts down Evals, Agent Builder and v1/prompts on 2026-11-30, three older image models on 2026-12-01 and six GPT-5 and o3 snapshots on 2026-12-11. The Claude Opus 4.5 row reads “not sooner than 2026-11-24”, which is a floor, not a shutdown date. Conferences: Microsoft Ignite from 11-17 to 11-20 and AWS re:Invent from 11-30 to 12-04, with GitHub Universe first on 10-28 and 10-29.
Updated 2026-11-11
One key across Claude, GPT and Chinese models, billed per token.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Four calendar lines that get misread
Sonnet 4.5's retirement date (not a floor)
The claude-sonnet-4-5-20250929 row now reads Deprecated: deprecated on 2026-09-30, retiring on 2026-11-30, replacement claude-sonnet-5-5. The row does not say “Not sooner than”, and the docs add that requests to models past the retirement date will fail.
Anthropic's minimum notice
Anthropic states at least 60 days' notice before retiring a publicly released model. As of 2026-10-26 the deprecation history still has no notice for Opus 4.5 or Haiku 4.5 and both rows still say Active, with Not sooner than November 24, 2026 and Not sooner than October 15, 2026 in the retirement column; a “not sooner than” floor is not a stop day, and the exact retirement dates have not been published.
The “at least through” on GPT-5.6 Sol pricing
The official sentence is “available until November 30, 2026”: the promotional price holds until that day; the vendor does not say it ends then. Pricing after 11-30 has not been published, and this page does not guess.
Conferences: dates and venues only
GitHub Universe on 10-28 and 10-29 (Fort Mason Center, San Francisco), Microsoft Ignite from 11-17 to 11-20 (Moscone Center, San Francisco), AWS re:Invent from 11-30 to 12-04 (Las Vegas). None of the three official pages names a model to be released there, so this page schedules no launches for them.
What this page carries and what it leaves out
As of 2026-10-26, on the official pages checked here, November has two dated model retirements: Claude Sonnet 4.5 on 2026-11-30 (Claude API) and gemini-3.1-flash-tts-preview on 2026-11-17 (Gemini API). The billing change is DeepSeek Search/Fetch charging per successful request from 00:00 Beijing time on 2026-11-01; the weights release is Kimi K3.1 Pro by 2026-11-04. Everything else is either an OpenAI product or legacy-model shutdown (11-30, 12-01, 12-11) or floor wording such as “not sooner than” and “at least through”. This calendar accepts three kinds of item that can be checked word for word on a vendor page: a stated date, a stated floor and a stated notice rule. Outlet predictions and social-media rumours stay off the calendar.
2026-10-01: OpenAI announces two new shutdown batches (2027-01-06 and 2027-04-01)
Re-checked on 2026-10-26, November gains three officially sourced dates: ① DeepSeek's official Search/Fetch docs state that these calls through Harness, Claude Code or Codex are not billed yet and will be charged per request from 00:00 Beijing time on November 1, 2026; the English price table lists $0.75 per 1,000 successful requests. ② Kimi's official blog says the full K3.1 Pro weights will be released by 2026-11-04. ③ Google's Gemini API deprecations page lists 2026-11-17 as the shutdown date for gemini-3.1-flash-tts-preview. The earlier OpenAI notice of 2026-10-01 still stands: gpt-5.3-codex, gpt-5.1 and gpt-5.4-nano leave the API on 2027-04-01 with six months' notice; tts-1, tts-1-hd and two gpt-4o-mini-tts snapshots shut down on 2027-01-06, replacement gpt-realtime-2.1-mini.
Three dates from November to early December
The “until” date for GPT-5.6 Sol's promotional pricing. The official changelog says GPT-5.6 Sol now costs $4 per million input tokens and $20 per million output tokens (the vendor calls it 20% lower on input and 33% lower on output), and the pricing-page footnote says the promotional pricing is available until November 30, 2026. The vendor gives only this “until” date, no end date; pricing after it has not been published.
Three things on one day: claude-sonnet-4-5-20250929 retires from the Claude API (notified 2026-09-30, replacement claude-sonnet-5-5); OpenAI shuts down the Evals dashboard and API (existing evals turn read-only on 2026-10-31), Agent Builder, and the v1/prompts API with reusable prompt objects; and AWS re:Invent opens, running until 12-04.
OpenAI shuts down gpt-image-1-mini, gpt-image-1.5 and chatgpt-image-latest, replacement gpt-image-2.5-sunburst or gpt-image-2.5-flare. Ten days later, on 2026-12-11, six snapshots follow: gpt-5-2025-08-07, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-5-pro-2025-10-06, o3-2025-04-16 and o3-pro-2025-06-10.
Layers: fixed date / floor / unpublished
Confirmed (verbatim on the official pages)
Anthropic (re-checked 2026-10-26): claude-sonnet-4-5-20250929 is Deprecated, notified 2026-09-30, retiring 2026-11-30, recommended replacement claude-sonnet-5-5; claude-opus-4-5-20251101 and claude-haiku-4-5-20251001 are still Active, with retirement not sooner than 2026-11-24 and 2026-10-15 respectively; publicly released models get 60 days' notice; these dates apply to the Claude API, Claude Platform on AWS, Claude models on Google Cloud operated by Anthropic, and Microsoft Foundry. OpenAI: GPT-5.6 Sol promotional pricing until 2026-11-30; Evals, Agent Builder and v1/prompts shut down on 11-30; three older image models on 12-01; six GPT-5 and o3 snapshots on 12-11; gpt-5.3-codex, gpt-5.1 and gpt-5.4-nano on 2027-04-01. DeepSeek: Search and Fetch cost $0.75 per 1,000 successful requests, billed per request from 00:00 Beijing time on 2026-11-01. Kimi: full K3.1 Pro weights by 2026-11-04. Google: gemini-3.1-flash-tts-preview shuts down on 2026-11-17. Conferences: Ignite from 2026-11-17 to 11-20 at Moscone Center, San Francisco (online 11-17 to 11-19), re:Invent from 11-30 to 12-04 in Las Vegas, GitHub Universe on 10-28 and 10-29 in San Francisco.
Unverified and unpublished (no rumours)
As of 2026-10-26 the vendors have not filled these fields, so they stay blank. First, exact retirement dates for Opus 4.5 and Haiku 4.5: the table shows only floors and both are still Active. Second, GPT-5.6 Sol pricing after 2026-11-30. Third, which models the three conferences will bring: the official pages give dates and venues but name no model. Fourth, Claude retirement dates on Amazon Bedrock and Google Cloud: Anthropic says those platforms set their own schedules, so the Anthropic dates here do not apply there. The DeepSeek and Kimi documents also give no November release date for a new model, and “coming in November” claims circulating online are not carried.
The two vendors date retirements differently
Anthropic: floor first, a date only after notice
Active models carry only a “Not sooner than” floor; the actual retirement date appears once the status switches to Deprecated, with at least 60 days' notice for publicly released models. Sonnet 4.5 followed that path: notified 2026-09-30, retiring 2026-11-30, 61 days apart counted from the two official dates. So when planning a migration, watch the status column, not the floor.
OpenAI: the notice carries the date, replacement on the right
A model deprecation notice comes with a concrete shutdown date, and the right-hand column usually names the replacement id. Unless safety or compliance concerns require a faster timeline, the stated minimum notice has three tiers: at least 6 months for generally available models, at least 3 months for specialized variants such as the Codex ones, and possibly as little as about 2 weeks for models with preview in the name. Both the 12-01 and the 12-11 batches were announced in June 2026.
Three things to do before 11-30
First, audit usage: Anthropic's own instructions are to open the Usage page in Claude Console and click Export; the CSV breaks usage down by API key and model, which shows the calls still naming claude-sonnet-4-5-20250929. Second, switch and re-test: set model to the recommended claude-sonnet-5-5 and, as Anthropic advises, test thoroughly well before the retirement date, because requests to the old id will fail after it. Third, the OpenAI side: if you use v1/prompts, move reusable prompt content into your application code; Evals users follow the official path to Promptfoo; Agent Builder users move to the Agents SDK or ChatGPT Workspace Agents; models shutting down on 12-01 and 12-11 get the replacement ids from the right-hand column of the deprecations page.
The QCode side: what these dates mean for calls here
As of 2026-10-26 QCode's /models marks claude-sonnet-4-5-20250929 as “Retired”, while claude-sonnet-5-5 is callable; switching changes only the model field, the key stays the same. Claude models use the Anthropic protocol with ANTHROPIC_BASE_URL set to https://api.qcode.cc/api; GPT models use the OpenAI protocol with base_url set to https://api.qcode.cc/openai/v1; one API key works on both protocols. Among the replacements for the 12-11 snapshots, gpt-5.6-sol and gpt-5.6-terra are callable on QCode, while QCode does not currently support gpt-5.6-luna; of the 2027-04-01 replacements, gpt-6-sol is callable, while gpt-6-luna is not currently supported either, so please use another GPT model, such as gpt-6.1-sol, gpt-6-sol or gpt-5.6-terra. QCode does not provide Gemini series; since 2026-11-11 QCode's /models lists kimi-k3.1-pro. Billing is per token; per-model prices are on /models. To try the replacement for a few days first: Not ready for a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 days; if you already have a Personal Plan, add a booster pack when you need more quota.
Frequently asked questions
Which AI dates in November 2026 are officially confirmed?
As of 2026-10-26 these dates are officially confirmed: DeepSeek Search/Fetch billing starts at 00:00 Beijing time on 2026-11-01; full Kimi K3.1 Pro weights by 2026-11-04; gemini-3.1-flash-tts-preview shuts down on 2026-11-17; GPT-5.6 Sol promotional pricing runs until 2026-11-30; Claude Sonnet 4.5 retires from the Claude API on 2026-11-30, the same day OpenAI shuts down Evals, Agent Builder and v1/prompts; 2026-12-01 and 2026-12-11, when OpenAI shuts down older image models and then GPT-5 and o3 snapshots. The conferences are Ignite from 11-17 to 11-20 and re:Invent from 11-30 to 12-04. Opus 4.5's 2026-11-24 is only a “not sooner than” floor.
When does Claude Sonnet 4.5 retire, and what replaces it?
On 2026-11-30, on the Claude API. The official deprecations table marks claude-sonnet-4-5-20250929 as Deprecated (notified 2026-09-30) and recommends claude-sonnet-5-5, whose own retirement is not sooner than 2027-09-28. Requests past the retirement date will fail. The date applies to the Claude API, Claude Platform on AWS and Microsoft Foundry; Amazon Bedrock and Google Cloud set their own schedules.
Will Claude Opus 4.5 stop on November 24?
Anthropic does not say that. The claude-opus-4-5-20251101 row is Active and its retirement cell reads “Not sooner than November 24, 2026”, a floor rather than a stop day. Anthropic promises at least 60 days' notice, and as of 2026-10-07 the deprecation history has no such notice; counted from that rule, even a notice sent on 2026-10-07 means no retirement before 2026-12-06. The same goes for Haiku 4.5 (not sooner than 2026-10-15).
Does GPT-5.6 Sol's promotional pricing end on November 30?
Not necessarily. The official sentence is “GPT-5.6 Sol's promotional pricing is available until November 30, 2026”: it lasts until that day and says nothing about ending on it. The official changelog gives the current price as $4 per million input tokens and $20 per million output tokens. Pricing after 11-30 has not been published.
Which OpenAI models shut down in early December, and what replaces them?
Two batches. On 2026-12-01: gpt-image-1-mini, gpt-image-1.5 and chatgpt-image-latest, replaced by gpt-image-2.5-sunburst or gpt-image-2.5-flare. On 2026-12-11: gpt-5-2025-08-07 and o3-2025-04-16 move to gpt-5.6-sol, gpt-5-pro-2025-10-06 and o3-pro-2025-06-10 to gpt-5.6-sol with reasoning.mode: pro, gpt-5-mini-2025-08-07 to gpt-5.6-terra, and gpt-5-nano-2025-08-07 to gpt-5.6-luna.
Will re:Invent, Ignite or Universe bring new models?
This page does not predict. None of the three official pages names a model; this page lists only dates, venues and registration details: GitHub Universe on 2026-10-28 and 10-29 at Fort Mason Center, San Francisco; Microsoft Ignite from 2026-11-17 to 11-20 at Moscone Center, San Francisco (online 11-17 to 11-19, in-person registration closes on 11-13 at 5:00 PM Pacific Time); AWS re:Invent from 2026-11-30 to 12-04 in Las Vegas. What actually launches is whatever the official pages say on the day.
Sources
Anthropic's official model deprecations page (status table, deprecation history, notice rule and the platforms the dates apply to); OpenAI's developer docs: the pricing page (GPT-5.6 Sol promotional-pricing footnote) and the deprecations page (the 2026-11-30, 12-01, 12-11 and 2027-04-01 batches); DeepSeek's official Search/Fetch docs (the billing section, prices and the 2026-11-01 start in English and Chinese); Kimi's official K3.1 Pro blog (weights by 2026-11-04); Google's Gemini API deprecations page (gemini-3.1-flash-tts-preview shuts down on 2026-11-17); the official AWS re:Invent 2026 page, the official Microsoft Ignite site and GitHub's official events page. QCode's connection settings come from the endpoints and API paths page on docs.qcode.cc, and the /models status from qcode.cc/models. All fetched on 2026-10-26.
The replacement models are callable today
claude-sonnet-5-5, claude-opus-5-5, gpt-6.1-sol, gpt-6-sol and gpt-5.6-sol share one QCode key, billed per token, with per-model prices on /models (as of 2026-10-26).
Related reading
October 2026 model release calendar
The previous edition: October's dated releases and shutdowns, hedge-only items, and where Haiku 4.5's “not sooner than” floor comes from.
Claude Sonnet 4.5 API retirement: the API and Copilot calendars
Same claude-sonnet-4-5-20250929, two removal calendars: Anthropic's API and GitHub Copilot are tracked separately, not merged into one deadline.
2026 AI Model Radar
Release and availability status of every 2026 frontier model on one page.
Crypto payment guide
Nine coins, 56 networks, zero platform fee for AI coding API credits.
Dates and quotes on this page were checked on 2026-10-26; the official pages of Anthropic, OpenAI, DeepSeek, Kimi, Google, AWS, Microsoft and GitHub take precedence, and the vendors may change them without notice. “Not sooner than” and “at least through” are floor wordings and are not rewritten as fixed dates here; dates marked as counted are derived from an official rule, not published by the vendor. Model availability is whatever /models shows. QCode is not affiliated with any of these vendors.