AI Mindset · Model Cheatsheets
Google Gemini

The Google-Native Agent That Can See and Act

Gemini 3.6 Flash reached general availability on July 21, 2026 and is Google’s current default fast model, alongside a cheaper Gemini 3.5 Flash-Lite tier. Build on an explicit model ID such as gemini-3.6-flash, never on the gemini-flash-latest alias. Two shutdown dates now matter more than any launch: the whole Imagen 4 family stops serving on August 17, 2026, and the whole Gemini 2.5 family follows on October 16, 2026. Google is connecting that intelligence to Search, Workspace, Android, Maps, multimodal creation and native computer use—turning its information ecosystem into an action layer.

Verified August 3, 2026Gemini 3.6 Flash · GA July 21Imagen 4 shuts down August 17Spark excludes the EEA and UKPin the model IDSearch · Workspace · Android
1 / Meet Gemini

One intelligence layer across Google’s working surfaces

Gemini is not one chat product. It is a family of models and experiences spanning Search, the Gemini app, Workspace, Android, developer tools and creative systems.

Personality

Fast, multimodal and connected

Gemini is strongest when the work combines current information, visual or audio context and a Google service that can help finish the task.

  • Search and Maps grounding
  • Text, image, voice and video
  • Consumer, enterprise and developer surfaces
Deploy it for

Information-rich action

Use Gemini for current research, Workspace creation, Android assistance, proactive monitoring and agents that need to see and operate interfaces.

  • Research and generative UI
  • Gmail, Docs, Slides and Sheets
  • Browser, mobile and desktop automation
Choose Gemini when

The context already lives with Google

Gemini has a structural advantage when the task begins in Search, Gmail, Drive, Android, Maps, YouTube or a Google developer environment.

  • Less context transfer
  • Live information
  • One model family across many surfaces

Where should Gemini meet the work?

Choose the surface that already owns the context.

Everyday assistant
Gemini app

Use the app for conversation, connected services, multimodal help, Daily Brief and emerging proactive-agent experiences.

  • Confirm the in-app default model before you standardize on it
  • Voice and camera context
  • Personal connected-app workflows
2 / What’s Current

A three-model July release resets the default—and Pro still has not shipped

On July 21, 2026 Google took Gemini 3.6 Flash and Gemini 3.5 Flash-Lite to general availability and described a security-tuned Gemini 3.5 Flash Cyber that only a pilot group can use. Gemini 3.5 Pro remains unavailable, and its rollout window has slipped. The more urgent news is at the other end of the lifecycle: Google’s deprecations page now dates the shutdown of the whole Imagen 4 family to August 17, 2026 and the whole Gemini 2.5 family to October 16, 2026.

Jul 21
Gemini 3.6 Flash and Gemini 3.5 Flash-Lite reached general availability
950M
Monthly Gemini app users, per Alphabet’s Q2 2026 CEO remarks on July 22, 2026
1B+
Monthly users of AI Mode in Search, per the same July 22 remarks
22B
Tokens per minute across Google’s model APIs, as reported by Alphabet on July 22
Aug 17
Shutdown date for every Imagen 4 model; the Gemini 2.5 family follows on October 16, 2026

Gemini 3.6 Flash GA July 21

Google does not call this a flagship. Its launch post calls it “our workhorse model that delivers better coding, knowledge work, and multimodal performance,” and the model page says it “balances speed with intelligence.” Status: Stable. On token efficiency Google’s own attribution is explicit: “According to the Artificial Analysis Index, it reduces output token usage by 17% compared to 3.5 Flash.” List price is $1.50 per 1M input tokens and $7.50 per 1M output tokens, unchanged since general availability on July 21, 2026.

  • Pin gemini-3.6-flash
  • The 17 percent figure is the Artificial Analysis Index, cited by Google — not a Google benchmark
  • Say “workhorse,” not “flagship”: Google is not claiming a capability ceiling
  • Re-baseline cost against your own traces

Gemini 3.5 Flash-Lite GA July 21

The high-throughput, low-cost tier. Google reports roughly 350 output tokens per second. List price is $0.30 per 1M input tokens and $2.50 per 1M output tokens.

  • Pin gemini-3.5-flash-lite
  • Volume classification and extraction
  • Test quality before you route traffic

Gemini 3.5 Flash Still stable

The May 19 model has not been withdrawn. July 21 added a new current default; it did not remove the model your existing integrations were built against.

  • No forced migration announced
  • Compare cost and output length
  • Move deliberately, not reflexively

Gemini 3.5 Flash Cyber Limited pilot

A security-tuned model for finding, validating and patching vulnerabilities. Access is a limited pilot for governments and trusted partners through CodeMender.

  • Not a purchasable capability
  • Do not present it as available
  • Track it as a signal, not a plan

Gemini 3.5 Pro Still coming

On May 19 Google said Pro was in internal use and would roll out the following month. On July 21 Google said it is testing with partners and will be made broadly available as soon as it is ready. It is absent from the API model list.

  • The timeline has slipped
  • Do not promise availability
  • Keep Flash as the current default

Computer use Announced for 3.5 Flash

Announced June 24, 2026 for Gemini 3.5 Flash: that model can see, reason and act across browser, mobile and desktop environments. Since July 21 the current default fast model is Gemini 3.6 Flash, and Google has published no statement extending computer use to it. Confirm coverage for whichever model ID you pin before you assume it carried over.

  • Announced on Gemini 3.5 Flash
  • Not confirmed on Gemini 3.6 Flash
  • Check the model ID you actually pin
Shutdown dateModelReplacement named by Google
August 17, 2026imagen-4.0-generate-001gemini-3.1-flash-image
August 17, 2026imagen-4.0-ultra-generate-001gemini-3.1-flash-image
August 17, 2026imagen-4.0-fast-generate-001gemini-3.1-flash-image
October 16, 2026gemini-2.5-progemini-3.1-pro-preview
October 16, 2026gemini-2.5-flashgemini-3.5-flash
October 16, 2026gemini-2.5-flash-litegemini-3.1-flash-lite

How much action should Gemini take?

Move from information to action only as permissions and consequence allow.

ObserveAct
Prepare
Draft the next step

Gemini prepares the email, plan, document, route or interface but does not change external state.

  • Useful default for work
  • Easy human review
  • Keep assumptions visible
3 / The Stack

Gemini’s advantage is the stack around the model

Each surface contributes a different kind of context: Search knows the live web, Workspace knows the work, Android knows the moment and the API turns the model into a building block.

Assistant

Gemini app

Conversation, connected apps, multimodal input, Daily Brief and emerging personal agents.

Information

Search AI Mode

Live web grounding, information agents, Maps and custom generative interfaces.

Work

Google Workspace

Gmail, Docs, Slides, Sheets, Meet and organizational content.

Research

Gemini Notebook

Source-grounded notebooks, audio and video overviews, reports and study tools. Renamed from NotebookLM on July 16, 2026, with a secure cloud computer for code execution and sync with the Gemini app.

Development

Gemini API

Model access, built-in tools, computer use and application-specific agents.

Coding

Antigravity

Agent-first development with artifacts, plans, screenshots and execution evidence.

Device

Android

Camera, voice, notifications, connected apps and contextual assistance close to the moment of work. At Galaxy Unpacked on July 22, 2026 Google said Gemini Intelligence task automation now reaches more than 40 apps, with Gemini Notebook preinstalled on the Galaxy Z Fold8 and Gemini on the Galaxy Watch 9.

Creative

Omni, Veo and image tools

Image and video creation connected to Gemini reasoning and the wider Google ecosystem.

July 31 drop

The July Gemini Drop

The monthly drop adds macOS voice input, an avatar feature, Dropbox, Zillow Rentals and Viator integrations, and personalized image generation in the United States only. It is also where Spark’s worldwide claim and its EEA, UK, Switzerland and Nigeria exclusion appear.

SurfaceContext it ownsBest outcomeAvailability caution
Gemini appPersonal conversation and connected appsEveryday help and proactive assistanceFeatures vary by plan, language and region
Search AI ModeLive web, shopping and MapsGrounded answer or interactive interfaceGenerative UI and agents roll out over time
WorkspaceOrganizational mail, files and meetingsNative work artifactEdition and admin policy matter
Gemini NotebookCurated user source setSource-grounded learning and briefingThe cloud computer is on AI Ultra and Workspace Expanded Access; Pro on the web is promised in coming weeks
API / AntigravityDeveloper-defined tools and environmentCustom agent or software workflowRequires safety and permission design
AndroidDevice, camera and contextual momentPersonal assistance and actionHardware and rollout vary
4 / Agentic Work

Google is turning Search and the device into persistent workers

The Gemini 3.5 and 3.6 line is designed to sustain work rather than only answer prompts. Spark, background information agents and computer use show Google’s attempt to make intelligence persistent across time and interfaces—while most of those surfaces are still limited, regionally excluded or unconfirmed.

A safe computer-use workflow

Computer use expands capability and the attack surface at the same time.

Define
Constrain the environment

Name the application, account, data, actions and stopping conditions that belong to the task.

  • Least-privilege access
  • No implicit cross-account reach
  • Explicit destructive-action rule

Gemini Spark Not in the UK or EEA

A personal agent intended to help across the digital day. The July Gemini Drop of July 31, 2026 frames the release as “Gemini Spark is going global... now available worldwide” — and the same page states: “Gemini Spark is not available in European Economic Area, United Kingdom, Switzerland, and Nigeria.” Read the exclusion list, not the headline.

  • “Worldwide” with four named exclusions
  • Unavailable to readers in the UK, the EEA, Switzerland and Nigeria
  • Do not build a UK or EU workflow on it

Managed agents July 28

Google: “Managed agents are now available on free tier projects. Developers can experiment with agentic workflows using an API key from a project without active billing.” The same post says the antigravity-preview-05-2026 agent now runs Gemini 3.6 Flash by default — that is one specific agent, not a platform-wide default — and notes model selection across 3.5 Flash and 3.5 Flash-Lite.

  • No active billing needed to experiment
  • The 3.6 Flash default belongs to one preview agent
  • Check which model each agent actually runs

Information agents in Search

Agents can monitor ongoing information needs and report changes with links for deeper action.

  • Repeated research
  • Alerts and monitoring
  • Source review remains necessary

Computer use in 3.5 Flash

Developers can build agents that operate browser, mobile and desktop interfaces. The June 24 announcement names Gemini 3.5 Flash; nothing published since says the capability extends to Gemini 3.6 Flash, so verify before you route an agent to the newer model.

  • Long-horizon automation
  • Verify the model ID carries it
  • Prompt injection and action risk

Daily Brief

The Gemini app can bring together timely personal information into a proactive morning view.

  • Reduce manual checking
  • Depends on connected context
  • Review privacy and relevance

Generative interfaces

Search can create a purpose-built interface or simulation instead of forcing every answer into prose. Google promised this for everyone in Search “this summer” on May 19, 2026; no later post confirms that it shipped.

  • Promised, not confirmed
  • Interactive decision support
  • Generated logic still needs testing

Android Halo

A status-bar surface that shows agent activity, announced May 19, 2026 in preview and described as rolling out later this year. It is a visibility surface, not a management console, and there is no July update.

  • Agent activity, not agent control
  • Announced in preview
  • Rollout status matters
5 / Creation

Gemini is joining reasoning and media production

Google’s creative stack matters because it can connect a source-rich research process directly to image, video and presentation outputs—without reducing creativity to a single model name.

Gemini Omni New

A multimodal model designed to combine reasoning with high-quality creation across text, image and video.

  • Cross-modal briefs
  • Cinematic output
  • Use the same context that shaped the idea

Nano Banana family

Google’s image-generation tools support rapid visual iteration and personalized creation across Gemini experiences.

  • Image generation and editing
  • Fast creative exploration
  • Check text, likeness and rights

Veo and Flow

Google’s video stack supports cinematic clips and scene-oriented workflows beyond a single generated shot.

  • Storyboard the intent
  • Use references and continuity
  • Plan for post-production

Slides and Workspace

Gemini can help move research and narrative into presentation form inside the same organizational environment.

  • Use brand templates
  • Link claims to sources
  • Inspect layout and story

Gemini Notebook briefing outputs

Audio, video, reports, quizzes and other outputs turn a curated source set into multiple learning formats. Google reports more than 30 million users and over 600,000 organizations on the product.

  • Source-grounded transformation
  • Useful for enablement
  • Keep audience needs explicit

Google Vids July 16

Vids gained Gemini Omni clip generation and personal avatars built from a selfie and a voice sample, watermarked with SynthID. Google lists it for Google AI Pro and Ultra plus Workspace business editions.

  • Check plan eligibility first
  • Get written consent for any likeness
  • Disclose synthetic presenters

Generative UI

Custom charts, tools and simulations are a new kind of creative output for questions that need interaction. Google promised general Search availability “this summer” on May 19; treat it as promised, not shipped.

  • Explain complex systems
  • Let users explore variables
  • Test the generated behavior
Multimodal campaign brief
Using the research notebook and brand system, create three campaign territories. For each, provide the strategic idea, hero image direction, six-second video beat and the evidence that makes the concept credible.
Source-grounded presentation
Turn this Gemini Notebook source set into an executive presentation. Keep one claim per slide, cite the exact source behind each claim and use our approved Slides theme.
Generative decision tool
Build an interactive comparison tool for these options using current Search and Maps data. Let the user change budget, distance and priority, and show how the recommendation changes.
Creative truth review
Inspect every generated visual for inaccurate text, impossible product details, misleading data or implied claims that the source material does not support.
6 / Behavior Playbook

Nine habits for using Gemini as an ecosystem

The winning behavior is choosing the right context surface, pinning what you build on and preserving the evidence as work moves toward action.

01

Start in the context owner

Use Search, Workspace, Gemini Notebook, Android or the API according to where the evidence lives.

02

Ask for grounding

Use live Search and Maps when current external reality matters.

03

Curate when authority matters

Use Gemini Notebook or a controlled source set instead of the whole web.

04

Use the actual modality

Give Gemini the image, video, voice or screen state rather than describing it poorly.

05

Separate prepare from act

Let Gemini draft first; require approval before consequential external actions.

06

Name the surface and plan

Availability differs across app, Search, Workspace, API, region and subscription.

07

Preserve source links

Keep live information attached as it moves into a document, deck or decision.

08

Test generated interfaces

Interactive outputs and computer-use actions need functional verification.

09

Pin the model, not the alias

Name gemini-3.6-flash or another explicit ID. Aliases move on Google’s schedule, not yours.

7 / Watch Outs

An ecosystem this broad is easy to overstate

Gemini features arrive across products, plans and regions at different times. The most common publishing mistake is turning a limited preview into a universal capability statement.

Rollout fragmentation

Plan, region, language, hardware and Workspace edition can all change access.

Grounding errors

A cited Search result can still be misunderstood, stale or weaker than the claim.

Computer-use risk

Interfaces can contain malicious instructions, unexpected prompts and sensitive information.

Background-agent drift

Long-running monitoring needs a clear objective, notification policy and stop condition.

Personal versus work context

Connected apps and device context can cross boundaries employees or administrators did not intend.

Generated media truth

Visual quality does not guarantee accurate text, products, scenes or implied evidence.

Product-name churn

Build guidance around durable workflows, with dated model notes, rather than temporary launch labels. NotebookLM became Gemini Notebook on July 16, 2026.

Announced is not available

Spark, Gemini 3.5 Pro, Flash Cyber and generative UI must retain their actual rollout state. Spark is the sharpest case: announced as available worldwide on July 31, 2026 and excluded from the European Economic Area, the United Kingdom, Switzerland and Nigeria on the same page.

Dated shutdowns, not vague deprecation

Imagen 4 stops serving on August 17, 2026 and the Gemini 2.5 family on October 16, 2026. Check the deprecations page for the exact model IDs before you quote a lifespan, and re-check it — it is a living table.

Whose benchmark is it

Google attributes the 17 percent output-token reduction for 3.6 Flash to the Artificial Analysis Index. Repeating it as a Google measurement misstates the evidence. Keep the third-party attribution attached.

Alias drift

No first-party Google page documents which model gemini-flash-latest serves today. The last documented switch in the release notes was January 21, 2026, to gemini-3-flash-preview, and Google’s stated policy is a hot swap with two weeks of email notice. Pin an explicit model ID. The Grok guide documents a live instance of the same failure: grok-voice-latest reprices itself by 60 percent per minute of audio on August 5, 2026, with no action by the customer.

Generative UI is unconfirmed

The only dated first-party statement is May 19, 2026: it “will be available for everyone in Search this summer.” No July post confirms that it shipped, so do not describe it as available.

Slipped timelines

Gemini 3.5 Pro was described in May as rolling out the next month and in July as testing with partners. Re-check announced dates before you repeat them.

Claim typeBefore publishingSafe wording
Model defaultCheck the official model release page“Default in the Gemini app as of…”
Model identifierCheck the release notes for the last documented alias changeName the pinned model ID and its GA date, never an alias
New agentCheck plan, region and tester status“Rolling out to…” or “available to…”
Workspace featureCheck edition and admin controlsName the Workspace surface and eligibility
Computer useCheck API/enterprise access, tool limits and which model IDs support itDescribe the supported developer surface
Creative capabilityCheck model, product and output availabilitySeparate announced model from shipping user feature
Model lifespanCheck the deprecations page for a dated shutdown and the named replacementGive the shutdown date and the replacement model ID
Regional availabilityRead the exclusion list, not the launch framingName the excluded regions in the same sentence as the launch
8 / Sources

First-party evidence behind this guide

These dated Google announcements anchor the August 3 edition. Google’s model lineup moves across the app, Search and the API at different speeds, so a claim true of one surface is not automatically true of another. One entry is a living reference page rather than an article: the deprecations table is the only authoritative record of when a model stops serving, so it is cited here with the date it was read. One date in this guide is not covered by that list: the January 21, 2026 switch of gemini-flash-latest to gemini-3-flash-preview comes from Google’s API release notes, a page that is rewritten in place and therefore cannot be cited here as a stable article. A reader who needs to rely on that date should confirm it directly.

AI Mindset

Explore the model cheatsheets