Google Gemini — DeepMind Multimodal Model Family + Assistant + Workspace Layer

Google Gemini (gemini.google.com, Google DeepMind, Mountain View, replaced Bard Feb 2024, 900M+ MAU at I/O 2026) active 2026-08, no discontinuation. It is not just a ChatGPT clone — it is a tiered model family (Nano / Flash / Pro / Ultra) + consumer assistant app + Workspace/Gmail/Docs/Android/Chrome embedding + creative stack (Nano Banana image, Veo 3.1 video, Flow, Music) + agent surface (Spark, Project Mariner, Auto Browse) + NotebookLM research. Category ai-productivity-data; pipeline_stage none. Killer differentiator = 1M-token Pro / 2M-token Ultra context + native Google grounding (Search/Gmail/Drive/Scholar) + Veo 3.1 video-with-audio + Android-default assistant; not ChatGPT (GPT-5.6 ecosystem, stronger agentic/voice/prose), not Claude (better long-form prose/code/instruction-following, 200K ctx), not Perplexity (citation-first research, no Workspace embed), not NotebookLM (Gemini is the model; NotebookLM is a source-bound sub-product). 2026 traps: consumer pricing reshuffled 3× since May 2026 (Ultra $249.99→$99.99/$199.99, Plus $9.99→$4.99 Jun 2026, 3.6 Flash default free Jul 2026, 3.7 Flash behind Spark Pro+/Ultra only); free tier Pro access variable/throttled; Deep Think + Veo 3.1 1080p + Spark + Mariner locked to Ultra; default training-on-conversations (opt-out buried); Android OS-level access to SMS/WhatsApp/calls draws privacy fire; prose still more “exec-summary” than Claude/ChatGPT; agentic coding less dependable than Claude Code/Codex.

1. One-line positioning

Gemini = the Google-shaped AI surface: open gemini.google.com / iOS / Android / Chrome side-panel → chat in 40+ langs with Gemini 3.6 Flash (free default) or bump to 3.1 Pro / 3.7 Flash (via Spark) → paste 1M-token PDF/codebase → Deep Research writes cited report → NotebookLM (500 notebooks, 300 sources each, AI Pro+) for source-bound Q&A + Audio Overview → Gemini Live voice interrupt → Gems (custom mini-assistants) + Canvas (prompt→app/quiz/infographic) → Nano Banana 2 image gen/edit in-thread → Veo 3.1 text/image→8s clip with synced dialogue/audio (Lite on Pro, 1080p on Ultra) → Flow filmmaking suite → Workspace: draft in Gmail, summarize Drive, build Slides, analyze Sheets, Meet recaps → Android default assistant (replaces Assistant, reads screen/SMS w/ consent) → Chrome Auto Browse (AI Pro+ multi-step form-fill, US-only Aug 2026) → Spark agent (AI Pro+ 160+ countries, Mariner deeper on $199.99 Ultra) → API via ai.google.dev (3.7 Flash $0.75/$3.75, 3.1 Pro $2/$12 per 1M, >200K doubles). Fits ai-productivity-data; pipeline_stage none (chat→deep research→notebooklm→workspace→spark→api). For Google-stack workers, long-doc/codebase ingestion, video-with-audio gen, Android users — default; for privacy-strict client work, Claude; for agentic coding, Claude/ChatGPT.

2. Core features

Layer	Capability	Notes
Models	Nano (on-device Android AICore, function-calling v3), 3.6 Flash (free default, 1M ctx, $0.75/$3.75 API), 3.7 Flash (Spark-only, AA Index 56), 3.1 Pro (1M ctx, Deep Think off, $2/$12), 3.1 Ultra (2M ctx, Ultra-only)	Flash family fastest in field; Pro trails GPT-5.6/Claude Opus 5 on hard reasoning
Multimodal	Text/image/audio/video in-out, chart understanding, 1M-token doc ingest	Best image/chart/video understanding 2026
Research	Deep Research (autonomous multi-source report w/ citations), NotebookLM (300 src/notebook, Audio Overview, 500 notebooks AI Pro+)	Grounding = Google Search + your Drive/Gmail/Scholar
Creative	Nano Banana 2 (image gen/edit), Veo 3.1 (8s clip + dialogue + SFX, Lite Pro / 1080p Ultra), Flow (Whisk-style dir suite), Music AI	Veo 3.1 = best integrated audio video 2026
Assistant	Gemini Live (interrupt voice), Gems (preset personas), Canvas (prompt→web app/quiz), Android default replace-Assistant, Chrome side-panel + Auto Browse	OS-level on Android 12+; reads SMS/WhatsApp/calls w/ consent → privacy debate
Agents	Spark (AI Pro+ 160+ countries, Jul 2026), Project Mariner (web agent, $199.99 Ultra), Jules (async code agent), Gemini CLI, Workspace actions (draft/summarize/refactor)	Agentic depth < ChatGPT/Claude Code
Workspace	Gmail draft/summarize, Docs co-write, Sheets formula/reason, Slides build, Meet recap, Drive Q&A — folded into Workspace plans 2025+	No separate add-on since 2025
API/Dev	ai.google.dev, Vertex AI Enterprise Agent Platform, 3.7 Flash $0.75/$3.75, 3.1 Pro $2/$12 (>200K doubles), Flash-Lite $0.30/$2.50	Introductory Flash rates double Jan 1 2027

3. Pricing (2026-08, USD consumer — Google changed this 3× since May 2026, verify gemini.google/subscriptions)

Tier	Monthly	Storage	Model / caps
Free	$0	15 GB	Gemini 3.6 Flash default, variable 3.1 Pro access, Deep Research (capped), Live, Canvas, Gems, Nano Banana image, 100 AI credits/mo, no Veo 3.1 video, no Deep Think, no Spark
AI Plus	$4.99 (cut from $9.99 Jun 2026)	400 GB	2× free limits, capped 3.1 Pro, Omni Flash video, Daily Brief, 200 Flow credits
AI Pro	$19.99 (ex-“Advanced”)	5 TB (incl Google One)	Broad 3.1 Pro, Deep Research full, Veo 3.1 Lite, NotebookLM 500 notebooks, Spark (160+ countries), 1,000 AI credits, Workspace full, YouTube Premium Lite (some regions)
AI Ultra (5×)	$99.99 (cut from $249.99 May 2026)	20 TB	5× Pro limits, Deep Think, Spark, Veo 3.1 1080p, Nano Banana Pro, 25k credits, YouTube Premium
AI Ultra (20×)	$199.99	30 TB	20× Pro limits, Project Mariner exclusive, everything above + earliest access
Workspace	bundled	—	Gemini in Gmail/Docs/Sheets/Meet since 2025, no separate fee
API	pay-go	—	3.7 Flash $0.75/$3.75, 3.6 Flash $0.75/$3.75, 3.5 Flash $1.50/$9, 3.5 Flash-Lite $0.30/$2.50, 3.1 Pro $2/$12 (>200K $4/$18) per 1M, intro Flash rates 2× Jan 1 2027
Enterprise	Custom	—	Vertex AI Enterprise Agent Platform, DPA, data residency, no consumer-training

pricing_type = freemium-plus-four-consumer-tiers-plus-workspace-bundle-plus-api-paygo-plus-enterprise; starting_price = $0; free_tier = "yes-generous-but-throttled; 3.6 Flash default, variable 3.1 Pro (shrinks at peak), Deep Research capped, Live/Canvas/Gems/Nano Banana image/100 AI credits/15GB, no Veo video, no Deep Think, no Spark, trains-on-by-default-opt-out-buried"; pricing_note = "gemini-bard-replaced-feb-2024-deepmind-active-2026-no-shutdown; ultra-249.99-to-99.99-and-199.99-may-2026-io; plus-9.99-to-4.99-jun-2026-storage-400gb; 3.6-flash-free-default-jul-2026; 3.7-flash-spark-only-pro-ultra; deep-think-veo-1080p-mariner-ultra-exclusive; pro-19.99-includes-5tb-google-one-youtube-premium-lite-some-regions; api-flash-rates-intro-double-jan-2027; workspace-gemini-bundled-since-2025-no-addon"

4. Access Type

access_type: web-app-plus-ios-plus-android-default-assistant-plus-chrome-sidepanel-plus-api-plus-workspace-embed

access_display: gemini.google.com · 📱 iOS/Android native (Android default assistant replace) · 🧩 Chrome side-panel + Auto Browse (US Aug 2026) · 📝 Gmail/Docs/Sheets/Slides/Meet embed · 🔌 ai.google.dev / Vertex AI · Nano on-device Android AICore · no Linux native app, no offline desktop, Spark not in EEA/UK/CH/Nigeria

5. Reviews (2026)

Source	Read
CompareGen (weighted)	8.7/10 — “very good free value, Google users justify paid easily; lags ChatGPT ecosystem, Claude prose”
AI Tool Radar	4.3/5 perf, 4.1/5 acc — “Deep Think 30–60s latency worth it on hard tasks; falls behind GPT-5.4 nuance, Perplexity sourcing; most feature-rich free tier”
T-Minus / Aivario	“Gemini wins doc-ingest (1M ctx) + Google-native research; Claude wins code/long-form; ChatGPT wins agentic/voice/image; Perplexity wins cited discovery”
Trustpilot / privacy press	Android OS access to SMS/WhatsApp/calls + default train-on + buried opt-out = caution for confidential/client work

Praise: free 3.6 Flash + Deep Research + NotebookLM-lite is the most generous free tier in AI (beats ChatGPT free/GPT-5.6-mini caps), 1M Pro / 2M Ultra context eats 500-page PDFs and 30k-line repos other models choke on, Veo 3.1 video-with-synced-dialogue is category-leading integrated audio, Workspace embedding (Gmail/Docs/Sheets/Meet) is zero-friction if you already live in Google, Android default assistant + Chrome Auto Browse make it OS-native not tab-native, AI Pro $19.99 w/ 5TB Google One + YouTube Premium Lite is effectively cheaper than ChatGPT Plus + Google One separately, Spark agent (160+ countries Pro+) broader than ChatGPT agent rollout in some regions.

Gripes: consumer pricing reshuffled 3× in 4 months (Ultra $249→$99/$199, Plus $9.99→$4.99, 3.6 Flash default Jul, 3.7 Flash Spark-gated) → plan comparisons rot fast, Deep Think + Veo 1080p + Mariner Ultra-exclusive = $19.99→$99.99 cliff, free Pro access variable/throttled at peak, default trains-on-conversations (opt-out buried, weaker than Claude default-off), Android OS-level SMS/WhatsApp/call access draws GDPR/privacy fire, prose “exec-summary-ish” vs Claude/ChatGPT, agentic coding less dependable than Claude Code/Codex on multi-step self-correction, Spark not in EEA/UK/CH/Nigeria, Gemini Live voice slightly behind ChatGPT Voice, benchmark numbers vendor-reported.

6. Best for / Not for

Best for	Not for
Google Workspace daily users (Gmail/Docs/Drive/Meet)	Privacy-strict client/regulated work (Claude default-off-training safer)
Long-doc / codebase ingestion (1M Pro / 2M Ultra)	Hard autonomous coding agents (Claude Code / Codex)
Video-with-audio gen (Veo 3.1) + image edit in-thread	Pure prose/creative writing quality (Claude/ChatGPT better)
Android users wanting assistant replacement	Non-Google stack teams (Copilot for M365)
Researchers using NotebookLM + Deep Research + Scholar	Citation-first minimal-hallucination research (Perplexity)
Budget buyers: free 3.6 Flash covers 80% casual use	People who hate price-tier churn (Gemini moved 3× in 2026)

7. Competitors

Tool	Lane	Starts
ChatGPT (Plus/Pro)	Generalist + GPTs/Codex/Sora/Voice, strongest ecosystem	$20/mo
Claude (Pro/Max)	Prose/code/long-doc, default-off-training, 200K ctx	$20/mo
Perplexity	Citation-first research, not Workspace-embed	$0 / $20-mo
Microsoft Copilot	M365-embedded, not Android-default	$0 / $20-mo
NotebookLM	Source-bound research sub-product (Google-owned)	Free / $19.99 via Gemini
Mistral/Le Chat, Grok	Alt generalists	Free / $20-mo