Dossier checked 28 Sep 2026
Gemini on the free plan Gemini 4 Argon, Gemini 3.8 Flash and Gemini 3.7 Flash are not named for free users. The Access Gap, checked 3 Oct 2026.
Newest frontier release
Gemini 4 Argon (GA)
13 Oct 2026 · the lab’s announcement
Generally available on the paid Gemini API and to Google AI Pro and Ultra subscribers in the Gemini app, after the restricted start on 30 Sep
Flagship list price, per million tokens
$2.00 in $10.00 out
Gemini 4 Argon, Standard tier (this price to 30 Nov 2026; $4.00 in and $20.00 out from 1 Dec) · pricing page
Weights
Closed flagship, some open
Partial. Gemini is closed; the Gemma 4 family is open under Apache 2.0. Source
Where you can buy the flagship
- Gemini API and Google AI Studio
- Gemini Enterprise Agent Platform (formerly Vertex AI)
- Gemini Enterprise
- Gemini app
- Search AI Mode and Workspace
- OpenRouter
9 releases in the year to 30 Oct 2026, 7 the year before · 21 in all since 2023
The question we ask of every lab: is most of its value in the model, which gets cheaper fastest, or in something other work is built on? Our answer: in something other work is built on.
Google’s durable position is the distribution it already owns: Search, Android, Workspace, and the TPU stack the models run on. Gemini compounds because it ships into surfaces a billion people open every day.
What would change our readA rival captures a primary surface, the default assistant on phones or in the browser, or the TPU advantage stops translating into cheaper frontier training.
01 · What it runs on, and the money
Trains and serves on
Its own chips in its own buildings, the only lab here where that is true end to end. The current TPU generation is Ironwood; the eighth, TPU 8t for training and 8i for inference, was announced in April 2026 for later this year. Google has also begun selling TPUs into customers’ own datacentres, with most of that revenue expected in 2027. Source
What it depends on Not a chip vendor. Google designs its training silicon and owns its datacentres, so it depends on TSMC to make the chips and on design partners: Broadcom, MediaTek (reported) and, since July 2026, Marvell. Source
Biggest risk Alphabet discloses no AI revenue line, yet raised about $70B of new equity and debt in one quarter to fund its build-out while its quarterly free cash flow turned negative. Source
- Revenue
- AI revenue is not disclosed separately. Alphabet turned over $119.8B in the quarter to 30 Jun 2026, of which Google Cloud was $24.8B, up 82%. Models were processing about 22 billion API tokens a minute. Source
- People
- 198,933 across Alphabet (30 Jun 2026); Google DeepMind is not broken out Source
- Structure
- Alphabet Inc., a Delaware public company with three share classes. Larry Page and Sergey Brin hold about 53% of the votes through super-voting Class B shares. Source
NotesThe flagship here is a Flash model because that is what Google itself ranks highest: Gemini 3.1 Pro is still a preview, at $2 in and $12 out, and 3.5 Pro has not been released. The bars use the introductory price; at the $1.50 and $7.50 list price from 1 Jan 2027, input is 1.2 times a year ago and output is 25% lower. DeepMind headcount is not disclosed.
02 · Releases
data through 30 Oct 2026
- Google 19 Gemini 4 Argon (GA)13 Oct 2026
Each dot is a frontier release at its date. Click one to find it in the full log. Every lab on one chart · Which plans and apps its new models reached
The landmarks
- 13 Oct 2026 Gemini 4 Argon (GA) Latest Generally available on the paid Gemini API and to Google AI Pro and Ultra subscribers in the Gemini app, after the restricted start on 30 Sep
- 19 May 2026 Gemini 3.5 Flash I/O launch: available to everyone in the Gemini app and AI Mode, and generally available in the Gemini API. Gemini 3.5 Pro was announced for the next month and never shipped
- Nov 2025 Gemini 3 Pro Flagship launch, frontier reasoning
- Mar 2025 Gemini 2.5 Pro Thinking model, top of leaderboards
- 9 Apr 2024 Gemini 1.5 Pro Million-token context, MoE architecture; a waitlisted preview from 15 Feb, public preview in the Gemini API from 9 Apr
- Dec 2023 Gemini 1.0 Gemini Pro in Bard and Nano on Pixel; Ultra followed in February
The full release logall 21, newest first
- 2026
- 13 Oct 2026 Gemini 4 Argon (GA) Frontier Generally available on the paid Gemini API and to Google AI Pro and Ultra subscribers in the Gemini app, after the restricted start on 30 Sep Source
- 30 Sep 2026 Gemini 4 Argon Restricted at launch: trusted cyber defenders through Fairwind, plus early testers; generally available from 13 Oct (its own row) Source
- 2 Sep 2026 Gemini 3.8 Flash Frontier Flash tier, generally available from launch Source
- 13 Aug 2026 Gemini 3.7 Flash Frontier Flash tier, generally available from launch Source
- 21 Jul 2026 Gemini 3.6 Flash Frontier Flash-tier refresh; Flash-Lite adds built-in computer use. No Pro-tier model shipped this cycle Source
- 19 May 2026 Gemini 3.5 Flash Frontier I/O launch: available to everyone in the Gemini app and AI Mode, and generally available in the Gemini API. Gemini 3.5 Pro was announced for the next month and never shipped Source
- Feb 2026 Gemini 3.1 Pro Frontier Preview update
- 2025
- 17 Dec 2025 Gemini 3 Flash Frontier New app default tier, rolled out to everyone at launch Source
- Nov 2025 Gemini 3 Pro Frontier Flagship launch, frontier reasoning
- Aug 2025 Nano Banana (2.5 Flash Image) Viral image generation and editing
- 6 May 2025 Gemini 2.5 Pro (I/O edition) Frontier Updated Gemini 2.5 Pro preview, released under its own name with its own post Source
- 17 Apr 2025 Gemini 2.5 Flash Frontier Fast reasoning tier, in public preview in the Gemini API, AI Studio, Vertex AI and the Gemini app from 17 Apr; GA in June Source
- Mar 2025 Gemini 2.5 Pro Frontier Thinking model, top of leaderboards
- 5 Feb 2025 Gemini 2.0 Pro Frontier Experimental public preview in AI Studio and Vertex AI, and in Gemini Advanced Source
- 2024
- 19 Dec 2024 Gemini 2.0 Flash Thinking Frontier Experimental Flash-tier reasoning model, in public preview in AI Studio and the API Source
- Dec 2024 Gemini 2.0 Flash Frontier Agentic-era multimodal default
- 14 May 2024 Gemini 1.5 Flash Frontier Fast, lightweight tier, in public preview from launch Source
- 9 Apr 2024 Gemini 1.5 Pro Frontier Million-token context, MoE architecture; a waitlisted preview from 15 Feb, public preview in the Gemini API from 9 Apr Source
- 8 Feb 2024 Gemini Ultra 1.0 Frontier Google’s largest model, public in the paid Gemini Advanced plan as Bard became Gemini; announced in December, gated until then Source
- 2023
- Dec 2023 Gemini 1.0 Frontier Gemini Pro in Bard and Nano on Pixel; Ultra followed in February
- 10 May 2023 PaLM 2 (Bard) Frontier Bard moved to PaLM 2 and opened to everyone in more than 180 countries, its waitlist removed Source
03 · When to reach for them
practical read
- Massive context
- Reach for Gemini 3.8 Flash when you need about a million tokens of text, image, video, audio or PDF in one pass (Sep 2026).
- Reasoning and coding
- Pick Gemini 3.8 Flash, which Google calls its most intelligent Flash model, for agent and coding work (Nov 2026).
- In the Google stack
- Use it where the work already lives: 3.8 Flash runs in the Gemini app, Search AI Mode, Sheets and Gemini Enterprise (Sep 2026).
04 · Weights, across all ten
the flagship
05 · Essays that read Google
from the archive
06 · About Google
at a glance
- Founded
- DeepMind 2010; merged 2023
- HQ
- Mountain View / London
- Leadership
- Demis Hassabis
- Parent
- Alphabet
- Also builds
- Search, Workspace, TPUs
- Known for
- Gemini, AlphaFold
Google’s AI is led by Google DeepMind, formed in 2023 by merging DeepMind (founded 2010, London) with Google Brain. Headquartered in Mountain View with major research in London, it produces the Gemini multimodal family as closed-weight models, with the smaller Gemma weights open.