Jiberish

The Bench

Google and Gemini: models, plans, and our verdict

Google's Gemini models across Workspace, the Gemini app, and the API: where they stand, plan prices, data terms, and recent changes.

Kyle Del FranciaUpdated 15 min read
In this piece

Google on the bench

  • WatchGemini: Gemini 4 Argon and Carbon, plus the Flash models, across Workspace, the Gemini app, and the API.Strongest at: Google Workspace · Long context · Cloud regions
See every verdict on the bench

Google builds the Gemini models and ships them across its own products: the Gemini app, Google Workspace, the Gemini API, and its cloud agent platform. For most businesses, Gemini arrives through Workspace long before anyone opens a developer console. This page covers where Google stands, the models and products that exist today, what they cost, how data is handled, and what changed this autumn. We checked every source linked here on November 10, 2026, using Google's developer documentation, Google Cloud and Workspace help pages, plan pages, and announcements. Google renames products and changes limits often, so confirm what your own admin console shows before planning around any detail here.

Where it stands

Gemini is on Watch: promising, but not one we recommend standardizing on yet. That is our judgment from hands-on testing and Google's published documentation.

The case for Gemini is real and got stronger this month. Google now has two generally available Gemini 4 models: Argon, GA since October 13, and Carbon, its strongest model, GA on the paid API and in Gemini Enterprise since November 10 (changelog). Both have a context window of more than a million tokens, and Gemini 3.8 Flash and 3.5 Flash-Lite cover faster and cheaper work (models). Its biggest advantage is distribution. For a team on Google Workspace, Gemini sits inside Gmail, Docs, Sheets, Slides, Drive, Meet, and Chat, with controls in the admin console the team already manages (Workspace pricing), and Google's HIPAA page lists the Gemini app and Gemini in Workspace among the services its BAA covers (HIPAA).

What holds it back is a lineup still in motion. Skills in Workspace apps, delayed in October, began rolling out to Rapid Release domains on November 9, with Scheduled Release domains starting November 30; they are not yet in Sheets or Meet and support English only (skills in Workspace). Carbon reaches the consumer Gemini app only on AI Ultra and does not appear in Workspace's Gemini app limits, Gemini 3.5 Pro has not been released, and the Gemini agent announced on October 8 still has no published plans or price. Google also renames products often: the Vertex AI Platform is now the Gemini Enterprise Agent Platform (name changes), and NotebookLM is now Gemini Notebook (rename).

Gemini moves to Trial for us when its business features settle: skills fully rolled out in Workspace with the admin controls Google describes, agent features with published plans, prices, and data terms, and our own hands-on tests holding up on real work.

The lineup today

Google's models page and API changelog list the first six rows below as current text models; the last covers a model that is expected but not released. GA means Google labels the model generally available or stable.

ModelAPI IDStatus on November 10, 2026Notes
Gemini 4 Carbongemini-4-carbonGA since November 10Google's strongest model, aimed at hard coding, research, and cybersecurity work; 1,048,576-token context shared by input and output; paid tier only (model page, pricing)
Gemini 4 Argongemini-4-argonGA since October 13Google's efficient frontier model; 1,048,576-token context shared by input and output; not on the free tier (model page)
Gemini 3.8 Flashgemini-3.8-flashGA since September 2Google's most capable Flash model
Gemini 3.6 Flashgemini-3.6-flashGA since July 21Previous-generation Flash; on the Gemini API, gemini-3.5-flash now routes here
Gemini 3.5 Flash-Litegemini-3.5-flash-liteGA since July 21Low-cost model for high-volume tasks
Gemini 3.1 Progemini-3.1-pro-previewPreviewGoogle's current Pro model; listed only in the models page's endpoint table
Gemini 3.5 ProNone publishedNot releasedIn July, Google said it was testing with partners (July 21 post); DeepMind's page still says coming soon (Gemini Pro)

"Gemini 4" now means two available models. Google says Argon "remains the ideal choice" for most software engineering and knowledge work, and that Carbon uses more tokens than Argon to work through harder problems, which raises the cost of each task (Carbon launch). A separate variant for security work, gemini-4-carbon-cyber-permissive, is limited to Google's Fairwind Program allowlist (Agent Platform release notes). We found no released Gemini 3.5 Pro.

For video, Gemini Omni 1.2 Flash (gemini-omni-1.2-flash) has been generally available on the Gemini API and Agent Platform since November 5. It makes clips of 3 to 15 seconds with audio and supports conversational editing (Omni 1.2 Flash).

The products sit on different surfaces, each with its own terms.

SurfaceWhat it includesAvailability
Gemini in WorkspaceGemini in Gmail, Docs, Sheets, Slides, Drive, Meet, Chat, and Vids, plus skills and Workspace Studio flowsGmail and Vids on all Business plans; the other apps from Business Standard (Workspace pricing); skills rolling out from November 9 (skills in Workspace)
Gemini app and Gemini Notebook, work accountChat assistant, plus the research notebook renamed from NotebookLM on July 16 (rename)Included with Workspace; Google's limits table lists Argon only for the AI Expanded Access add-on and does not list Carbon (AI Expanded Access)
Gemini app, personal accountFree, Google AI Plus, Pro, and Ultra plansArgon for AI Pro and Ultra since October 13 (Gemini app post); Carbon rolling out to AI Ultra since November 10 (Carbon launch), in select countries (Gemini subscriptions)
Gemini for business (Gemini Enterprise app)Agent workspace with connectors to Workspace, Microsoft 365, Slack, and others; where Google presents the Gemini agentBusiness edition offers a 30-day self-serve trial; Carbon GA since November 10 in global, us, and eu, off by default in the web and mobile apps until an admin turns it on (release notes)
Gemini API and AI StudioDeveloper access to the models aboveFree and paid tiers; Carbon and Argon on the paid tier only
Gemini Enterprise Agent PlatformThe renamed Vertex AI (name changes)Argon in global, us, eu, europe-west2, and asia-southeast1 (Argon model card); Carbon GA since November 10 in global, us, and eu, though its model page describes access for "verified enterprise customers in select regions" (Carbon model card)

Plans and pricing

All figures are US list prices as of November 10, 2026, taken from Google's own pages.

Google Workspace planPer user per month, 1-year commitmentGemini features
Business Starter$7Gemini in Gmail and Vids, the Gemini app, basic Gemini Notebook, skills and Workspace Studio
Business Standard$14Adds Gemini in Docs, Sheets, Slides, Drive, Meet, and Chat, and expanded Gemini Notebook
Business Plus$22Same AI features as Standard, plus Vault
EnterpriseContact salesAdds controls such as data loss prevention and data regions

The Workspace pricing page also advertises a limited-time discount for new customers; the figures above are the standard prices. Business plans cap at 300 users. Video generation in Vids is capped at 200 seconds a month on Business Starter, 500 on Business and Enterprise Standard and Plus, and 2,000 with the AI Expanded Access add-on (AI Expanded Access). That add-on, which raises limits and adds Argon in the Gemini app at up to 200 prompts a day, has no published price; Google says pricing depends on currency and term.

Personal Google AI planPer monthNotes
Free$0Flash-Lite model in the Gemini app
Google AI Plus$4.99Flash model, 400 GB storage
Google AI Pro$19.99Argon and Pro models, 5 TB storage
Google AI UltraFrom $99.99$99.99 for 5x Pro's limits, $199.99 for 20x; Carbon rolling out since November 10

These come from Gemini subscriptions and carry consumer data terms. Gemini Enterprise Business edition starts at $21 per seat per month (Business edition).

API model, paid tierInput per 1M tokensOutput per 1M tokens
gemini-4-carbon$2.00 through Nov 30, then $4.00$10.00 through Nov 30, then $20.00
gemini-4-argon$2.00 through Nov 30, then $4.00$10.00 through Nov 30, then $20.00
gemini-3.8-flash$0.75 through Dec 31, then $1.50$3.75 through Dec 31, then $7.50
gemini-3.6-flash$0.75 through Dec 31, then $1.50$3.75 through Dec 31, then $7.50
gemini-3.5-flash-lite$0.30$2.50
gemini-3.1-pro-preview$2.00, or $4.00 over 200K tokens$12.00, or $18.00 over 200K tokens

Prices are from Gemini API pricing; output includes thinking tokens. Batch and Flex input and output cost half the standard rate for these models. Search grounding includes 5,000 free requests a month, shared across Gemini 3 and newer models, then costs $14 per 1,000. Gemini Omni 1.2 Flash is paid-tier only, at $1.50 per 1M input tokens and $12.00 per 1M video output tokens, which Google puts at $0.07 per second of 720p video, 30% below Omni 1.1 (Omni 1.2 Flash).

Data, privacy, and compliance

Account type decides the terms. For qualifying Workspace editions, Google's Workspace privacy hub says interactions stay within the organization, content is not human reviewed or used for model training outside your domain without permission, and Gemini follows existing Workspace permissions, including IRM limits on downloading or copying. Admins set retention: 90 days to indefinite for Gemini in Workspace, and up to 36 months for the Gemini app. Since October 22, Vault can export Gemini in Workspace conversations filtered by user, organizational unit, and date on Business Plus, Enterprise Standard and Plus, and some other editions (Vault update). The hub lists SOC 1, 2, and 3, ISO 9001, ISO/IEC 27001, 27701, 27017, 27018, and 42001 certifications, and FedRAMP High authorization.

Skills in Workspace are on by default for eligible editions. Admins can turn them off by domain, organizational unit, or group, external sharing follows the organization's Workspace Studio sharing controls, and skill activity appears in Workspace Studio log events (skills in Workspace).

Personal accounts work differently. With Keep Activity on, Google says it uses Gemini app activity to develop and improve its services, including training generative AI models; a subset of chats is read by human reviewers, and reviewed chats are kept for up to three years (Gemini Apps privacy hub). Company data belongs in a managed Workspace account.

On the API, Google uses free-tier content to improve its products, and human reviewers may read it. Paid-tier prompts and responses are not used to improve products and are logged for a limited period for abuse monitoring (Gemini API terms). Since October 26, new paid-tier projects store Generate Content requests by default, with logs kept 55 days unless you shorten the window or set store to false (logs and datasets).

For HIPAA, Google's compliance page lists the Gemini app (excluding Gemini in Chrome), Gemini in Google Workspace, and the Gemini Mac app among Workspace services in the Google Cloud BAA, in a list updated October 28, 2026. Gemini Enterprise, Agent Platform, and Gemini Notebook Enterprise appear on the Cloud list. The Workspace privacy hub says Gemini Notebook and Gemini in Chrome are not covered by the BAA, and Google's Workspace FAQ notes that some features may be blocked once a BAA is signed. Gemini for healthcare, which Google Cloud launched in preview on November 10, does not yet appear on the BAA list (Gemini for healthcare). Our health and wellness article covers what has to change before AI touches patient data.

What changed recently

  • August 13: gemini-3.7-flash reached GA; since October 8 it routes to 3.8 Flash (changelog).
  • September 2: Gemini 3.8 Flash reached GA on the API and in Gemini Enterprise, and came to the Gemini app for AI Pro and Ultra subscribers (announcement).
  • September 18: The Gemini API limited 2.5-model access to recent users and pointed new projects to 3.5 Flash-Lite or 3.8 Flash.
  • September 22 to 29: Meet note-taking added a one-page quick-notes summary and an admin option to take notes only in meetings of three or more (September updates).
  • September 30: Google introduced skills for the Gemini app and Workspace to replace Gems (skills) and began giving cyber defenders early access to Argon (Argon preview).
  • October 8: Google Cloud announced the Gemini agent, a single agent for knowledge work, content, and code (announcement).
  • October 13: Argon reached GA on the API, Agent Platform, and Gemini Enterprise, and came to the Gemini app for AI Pro and Ultra subscribers (Argon launch). The same day, Google delayed the Workspace skills launch (skills update).
  • October 15: Gemini Enterprise admins can turn on Anthropic's Claude Opus 5.5 and Sonnet 5.5 in the app, billed pay-as-you-go (release notes).
  • October 20: Agent Platform's lifecycle table lists this as the retirement date for gemini-2.5-pro, gemini-2.5-flash, and gemini-2.5-flash-lite (model versions).
  • October 28: Google announced Gemini 4 Carbon, starting with a small group of cyber defenders (announcement).
  • October 30: Gemini in Slides can edit a whole deck from one prompt on Business Standard and above (October updates). Computer Use became GA on the Gemini API (changelog).
  • November 2: The Gemini API deprecated gemini-2.5-pro and gemini-2.5-flash. From December 1, requests route to 3.8 Flash and 3.5 Flash-Lite respectively and bill at those models' rates; gemini-2.5-flash-lite is not deprecated (deprecations).
  • November 5: Gemini Omni 1.2 Flash reached GA on the Gemini API and Agent Platform, with clips up to 15 seconds and a 30% price cut (announcement). It came to Google Vids the same day on Business Starter through Plus, Enterprise Starter through Plus, and other editions (Vids update); Google's post says anyone with a Google account can make up to three 15-second videos a month in Vids at no cost (Vids post).
  • November 6: Gemini Enterprise replaced Veo 3.1 with Omni 1.2 Flash for video generation in the global region, and said bundled Gemini Code Assist entitlements end March 31, 2027 (release notes).
  • November 9: Skills in Workspace began rolling out to Rapid Release domains, invoked in Gmail, Docs, Slides, Drive, and Chat, on Business Starter through Plus, Enterprise Starter through Plus, and the AI Expanded Access add-on (skills in Workspace).
  • November 10: Gemini 4 Carbon reached GA for paying API developers and in Gemini Enterprise, and began rolling out to AI Ultra subscribers in the Gemini app, AI Mode, and Antigravity (Carbon launch). Gemini's thread summaries and Help me write in Gmail began expanding to all Workspace customers (Gmail update).

Where it fits

The natural fit is an organization standardized on Google Workspace, on Business Standard or higher, where Gemini adds no new vendor or sign-in. Business Starter includes Gemini in Gmail but not in Docs or Meet, so the edition belongs in the decision (Workspace pricing).

For developers, the Gemini API and the Agent Platform offer long context and a range of price points. Argon is generally available in several cloud regions, including the EU and Asia (Argon model card), while Carbon is limited to global, us, and eu (Carbon model card). Teams with data-location requirements should look at those regional options early. Because Carbon is priced the same per token as Argon but uses more tokens per task, start new work on Argon and move a task to Carbon only when Argon falls short.

Consumer and business terms differ sharply. Google says chats in the consumer Gemini app can be used to improve its models and reviewed by people (Gemini privacy), while Workspace keeps generative AI interactions within the customer's organization under its own terms (privacy hub). Company data belongs on business accounts.

What we're watching

  • The rest of the Workspace skills rollout: Scheduled Release domains start November 30, and skills are not yet in Sheets or Meet (skills in Workspace). On the Gems side, Gems move to Settings on November 17, 2026, and business accounts lose them no sooner than March 1, 2027 (skills).
  • Plan eligibility, price, and data terms for the Gemini agent. In Alphabet's October 28 earnings remarks, Sundar Pichai said agents will work across Google's products, without giving a date (Q3 remarks).
  • Carbon for Workspace customers, which Google's Gemini app limits table does not yet list, and a release of Gemini 3.5 Pro.
  • Price changes: Argon and Carbon rise to $4 and $20 per 1M tokens on December 1, and 3.8 Flash to $1.50 and $7.50 on January 1, 2027.
  • Retirements and routing: several preview audio models on the Gemini API have November 17, 2026 as their earliest shutdown date; gemini-3.6-flash retires from Agent Platform on November 19; gemini-2.5-pro and gemini-2.5-flash start routing to newer models on the Gemini API on December 1; and gemini-omni-1.1-flash routes to 1.2 Flash on December 3 (deprecations).
  • Gemini Notebook Enterprise web addresses change on November 16, which may require firewall and allowlist updates (release notes).
  • BAA coverage for Gemini for healthcare as it moves beyond its preview with launch partners.
  • Which model powers Workspace drafting. The pricing page offers Business Standard and above Google's most capable models without naming one; a named model would tell teams when drafting behavior may change.

From verdict to deployment

Putting Google to work is the hard part.

We design and build AI into the websites and applications a business runs on, automate workflows with agents that work across its tools, and roll AI out across whole organizations, with every team trained on its real work. We build on whichever lab fits the job, and we stay through launch and adoption.

Book a free call

Put this into practice

Bring the decision you’re working through.

A free discovery session is a chance to review your goal, the tools or website you have, and the help you need. We’ll identify a sensible next step and discuss scope before any project commitment.

Custom AI Systems & Agents →AI Advisory & Training →Book a free call