Updated
AI Pricing This Week: Claude Ultra, Gemini 2.5, Nov 2-8
Short answer. From November 2 to 8, 2026, no list price moved. Claude listed a new Ultra plan at $200 a month and Max at $100, with no Max 20x option (November 3). Google dated the shutdown of Gemini 2.5 Pro and 2.5 Flash for December 1, 2026 (read November 6). Mistral deprecated Medium 3.5 and Large 3, replaced by Large 4.
In plain words. No API price moved this week. Claude changed its plans: a new $200 plan, and the old top option is gone. Google set the day two older Gemini models stop answering. Mistral marked two models for removal and named one replacement for both.
Update, November 16, 2026: Mistral now dates Mistral Large 4’s general availability. Its model page, read on November 16, 2026, reads “November 5, 2026 GA Mistral Ecosystem Open Weight License v 26.10”, and its Markdown version “Release date: 2026-11-05”. Where this post says Mistral does not say when the stage changed, Mistral now gives November 5, 2026. The price is unchanged, $0.70 / $0.07 / $2.10.
This is the fourth weekly summary. It covers Monday, November 2 to Sunday, November 8, 2026. Every fact comes from AI Signal’s dataset, change log and deprecation records, with the provider page it was read on and the date. The third summary mentioned Mistral’s deprecations as read after its week closed; they are told in full here.
The week in three lines. Claude’s plans changed shape on November 3: Ultra at $200, Max at $100, Max 20x no longer listed. Google’s deprecations page, read on November 6, gave shutdown dates to three models and the documentation moved to a new address. Mistral deprecated two models on October 31 and moved Large 4 to general availability.
Shutdowns and deprecations
Mistral Medium 3.5 and Mistral Large 3, deprecated on October 31. The weekly check of November 2 was the first to find neither model on Mistral’s pricing page. On November 3 their model pages said “This model is deprecated. Use Mistral Large 4 for new integrations.”, “Deprecation date 10/31/2026” and “Replacement Mistral Large 4”, with no retirement date on the page as shown in a browser. Both rows are deprecated, with the last prices read on October 26: $1.50 / $7.50 for Medium 3.5 and $0.50 / $1.50 for Large 3. Mistral Large 4 costs $0.70 / $2.10, so the move is cheaper from Medium 3.5 (1.50 ÷ 0.70 = 2.14× on input) and dearer from Large 3 (0.70 ÷ 0.50 = 1.40×). The arithmetic is in Mistral deprecates Medium 3.5 and Large 3.
GLM 5.2 on Mistral, retired on October 31. Mistral’s changelog says: “Z.ai GLM 5.2 ( zai-glm-5-2 ) is deprecated and retires on October 31, 2026. Use Z.ai GLM 5.3 ( zai-glm-5-3 ) instead, at the same price.” We read that line on November 3 and marked GLM 5.2 as retired, with the $1.40 / $4.40 it had; GLM 5.3 is on the pricing page at the same figures.
Mistral Large 4, GA since the week of November 6. On October 30 the model’s page read “October 6, 2026 Public Preview Mistral Ecosystem Open Weight License v 26.10”; on November 6 it reads “October 6, 2026 GA Mistral Ecosystem Open Weight License v 26.10”. The price did not move: “Mistral Large 4 ↗ | $0.7 | $0.07 | $2.1 |”. Mistral does not say when the stage changed; the change was first seen on November 6.
Gemini 2.5 Pro and 2.5 Flash, December 1; Gemini Omni 1.1 Flash, December 3. Google’s deprecations page, read on November 6, prints “gemini-2.5-pro | June 17, 2025 | December 1, 2026 | gemini-3.8-flash ( gemini-2.5-pro is no longer available in Google AI Studio; effective December 1, 2026, API requests to gemini-2.5-pro will automatically route to gemini-3.8-flash ) |” and the same row for gemini-2.5-flash, routed to gemini-3.5-flash-lite; the pricing page says the requests “will automatically route to Gemini 3.8 Flash ( gemini-3.8-flash ) and be billed at Gemini 3.8 Flash rates”. What that costs, per million tokens:
| From | Its price | Routed to | Its price today |
|---|---|---|---|
| Gemini 2.5 Pro | $1.25 / $10.00 | Gemini 3.8 Flash | $0.75 / $3.75 through December 31, 2026; $1.50 / $7.50 from January 1, 2027 |
| Gemini 2.5 Flash | $0.30 / $2.50 | Gemini 3.5 Flash-Lite | $0.30 / $2.50 |
| Gemini Omni 1.1 Flash | $9.00 (text) / $17.50 (video) output | Gemini Omni 1.2 Flash | $9.00 (text) / $12.00 (video) output |
For 2.5 Pro the routed request is cheaper while the Flash promotion lasts (1.25 ÷ 0.75 = 1.67× on input) and dearer at the list price Google prints for January (1.50 ÷ 1.25 = 1.20×). For 2.5 Flash the figures are the same. Google does not say when it published the dates; AI Signal first saw them on November 6, and the reading of October 26 had none. The comparisons, request by request, are in Gemini 2.5 Pro and 2.5 Flash shut down on December 1. The Omni models are video models and stay out of the price table (below).
Price changes
No list price moved in the week. The only entry in the price-change log is ours: on November 6 Gemini Omni 1.2 Flash entered the dataset and was kept out of the table, the calculator and the counts from that day, as Gemini Omni 1.1 Flash has been since September 30, because Google’s model page describes “a high-performance model for fast, conversational video generation and editing”. The entry is marked as a change of our criterion, not a price change by Google.
New models
| Model | Released | Added here | Input / output per 1M tokens |
|---|---|---|---|
| GPT-6 Cyber | no date | November 6 | $20.00 / $100.00 |
| Gemini Omni 1.2 Flash | November 5 | November 6 | $1.50 / $9.00 (text), out of the table |
GPT-6 Cyber is sold with restricted access. Its model page, read on November 6, prints “Input $20.00”, “Cached input $2.00”, “Output $100.00”, “1,050,000 context window” and “Your organization needs Daybreak Red access and approval to use GPT-6-Cyber.” The Markdown version of OpenAI’s pricing page prints ”| gpt-6-cyber | $20.00 | $100.00 | $2.00 | $25.00 | $50.00 | - | - | - | - | - |”: a 30-minute cache write of $25.00 and a 12-hour one of $50.00, and no long-context figures. Its page on this site is kept out of the search index, as GPT-5.6 Cyber’s is.
gpt-6-astra-law is not a model row. The same pricing page has a block headed “Legal model configurations” with one row, “gpt-6-astra-law | $10.00 | $50.00 | $1.00 |”, followed by “GPT-6 Astra for Law promotional pricing is available at least through December 31, 2026.” That is GPT-6 Astra’s own short-context price, under a “Model configuration” column, with no model page of its own (404 on November 6). It is quoted in GPT-6 Astra’s row and gets no row of its own.
Gemini Omni 1.2 Flash was released on November 5, according to Google’s deprecations page (“gemini-omni-1.2-flash | November 5, 2026 | No shutdown date announced |”), at “Input price | Not available | $1.50 (text / image / video) |” and $9.00 per million output tokens for text, $12.00 for video. It has a row, with its prices and dates, and stays out of the table for the reason above.
Google’s documentation moved
Since at least November 3, every address under ai.google.dev/gemini-api/docs/ answers with a redirect to aistudio.google.com/docs/, and the new pages are drawn by JavaScript: a script receives only the title. From November 6 AI Signal reads each Google page from the Markdown source Google serves at the same address with “.md”, and every Google row cites the new address; the 15 priced rows were checked against the Standard tab that day and all matched. The weekly check of Monday, November 9 was the first to run on the new source, with no price mismatch. The methodology describes the reading.
Subscriptions and coding tools
Claude: Ultra at $200, Max at $100, Max 20x gone (November 3). On October 17 claude.com/pricing printed Max as “From $100 Per month” with “Choose 5x or 20x more usage than Pro*”. On November 3 it prints “Max For all-day work $100 Per month” and a new card, “Ultra For your longest tasks $200 Per month”, with “No 5-hour limit ‡” and “10x Pro usage every week*”; the Claude Code page, read with a browser the same day, shows Pro, Max ($100) and Ultra ($200) and names no 20x option. In Spain, read by hand on November 3 and again on November 6, the page prints “Max Para trabajar todo el día 90 € Por mes” and “Ultra Para tus tareas más largas 180 € Por mes”, with no 5x or 20x choice; on October 27 it printed Max as “Desde 90 € Por mes”. The dollar and euro figures existed before; what each buys is described differently, and the 20x option is no longer listed. When the change happened is not known: between October 26 and November 3. Claude’s plans keep Max 20x as a plan no longer listed, with its last reading; the post is Claude Ultra: the new $200 plan.
Google AI Pro and AI Ultra, re-read on November 3. The prices did not move: $19.99, $99.99 and $199.99 a month. The page no longer lists Jules among what the plans include; on October 17 it printed “Jules” with the footnote “Jules is an asynchronous coding agent that is currently in beta.” Google Gemini’s plans and the Jules page say so.
Gemini Code Assist: no longer sold to new customers since October 9 (seen November 6). The address this site read on October 19, codeassist.google/products/business, now serves a Google Cloud page that says “Starting October 9, 2026, you can no longer purchase new Gemini Code Assist Standard or Enterprise subscriptions.” and “If you have an annual subscription, you can continue to auto-renew your subscription until January 31, 2027.” The page gives no price. Gemini Code Assist keeps its last prices, read on October 19, and is no longer counted among the tools on sale on the AI coding tools page.
Kiro: credit pooling for teams (seen November 6). On November 3 kiro.dev/pricing said “For team usage, each developer needs their own subscription.”; on November 6 and 7 it prints “Credit pooling for teams of 2 or more”, “$20 per seat / month” and “500 included credits / seat / month”. The individual plans did not change. Kiro records the new plan and the change.
Replit and Bolt, re-read on November 3. No price moved: Replit Core $18 and Pro $90 a month; Bolt Pro $25 and Teams $30 a month, $18 and $27 on the yearly view. Bolt’s page no longer prints the Compare table or its tax footnote, “Plus applicable taxes”, that it printed on October 27, so Bolt’s tax is now recorded as not stated.
Posts of the week
- November 6: Claude Ultra: the new $200 plan, and Max 20x is gone.
- November 6: Gemini 2.5 Pro and 2.5 Flash shut down on December 1.
- November 6: Mistral deprecates Medium 3.5 and Large 3.
- November 7: the third weekly summary, October 26 to November 1.
After the week closed: on Monday, November 9, OpenAI’s pages gave GPT-5.6 Sol’s promotion an end date, November 30, 2026, where they had said “at least through November 21”; Together AI’s deprecations page listed Qwen3.8 Max as removed from serverless on November 4, so its offering is a retired record on open-model prices by host; GPT-6 Rosalind entered the dataset with restricted access; and Baseten’s pricing page, now drawn by JavaScript, is read with a headless browser. All of it belongs to the next summary.
What comes next
Dates as the providers print them, all on Deprecations:
- November 30: Anthropic retires Claude Sonnet 4.5: “November 30, 2026 | claude-sonnet-4-5-20250929 | claude-sonnet-5-5 |”. The same day is the last of Gemini 4 Argon’s introductory price, “$2.00 through November 30, 2026. $4.00 starting December 1, 2026.”, and of GPT-5.6 Sol’s promotional price as OpenAI’s page states it since November 9.
- December 1: Gemini 2.5 Pro routes to Gemini 3.8 Flash and Gemini 2.5 Flash to Gemini 3.5 Flash-Lite.
- December 3: Gemini Omni 1.1 Flash routes to Gemini Omni 1.2 Flash.
- December 11: OpenAI shuts down GPT-5, GPT-5 mini, GPT-5 nano, GPT-5 Pro, o3 and o3-pro.
How we checked
The facts were listed by our weekly script from AI Signal’s own data files: the models dataset, the price-change log, the deprecation records, the subscription and coding-tool data and the posts of the week. Every price and date was read on the provider’s own page, on the date given; Google’s pages were read from their Markdown source, the Claude Code page with a browser and Claude’s Spanish page by hand. Model prices are list prices in US dollars per million tokens. We do not know why any of these prices are what they are, and this summary does not guess.
Frequently asked questions
Which AI model prices changed between November 2 and 8, 2026?
No list price moved. AI Signal added GPT-6 Cyber at $20.00 / $100.00 per million tokens, with restricted access, and Gemini Omni 1.2 Flash at $1.50 / $9.00, kept out of the table as a video model; Mistral Large 4 moved from Public Preview to GA at its unchanged $0.70 / $2.10.
Which models were deprecated or given shutdown dates that week?
Google's deprecations page, read on November 6, 2026, dates Gemini 2.5 Pro and 2.5 Flash for December 1, 2026 and Gemini Omni 1.1 Flash for December 3. Mistral deprecated Medium 3.5 and Large 3 on October 31, pointing both to Large 4, and retired GLM 5.2 the same day, as read on November 3.
What changed in AI subscriptions and coding tools that week?
Claude lists a new Ultra plan at $200 a month, 180 € in Spain, and Max at $100, 90 € in Spain, with no Max 20x option, as seen on November 3. Google stopped selling Gemini Code Assist to new customers on October 9, seen on November 6, and Kiro added credit pooling for teams at $20 per seat a month.
Written with AI assistance from this site's own verified pricing dataset and reviewed by Álvaro Lucero Rufino, who is responsible for what stays published here. Every figure links to the page where it was verified, with the date it was read; if a number here and a number on that page ever disagree, the page is the one to trust. Automated publishing was discontinued in September 2026; posts that no longer met that standard were removed rather than left up.