Google

Google develops AI models through Google DeepMind and offers them via consumer products (Gemini app, Search), cloud services (Vertex AI), and open-source releases.

We're taking another big step on the path toward AGI. — Demis Hassabis, CEO of Google DeepMind

Text Models

Gemini (Flagship)

Google's multimodal flagship model family, powering the Gemini app, AI Mode in Search, and enterprise applications.

Gemini 3.1 Series (February 2026)

• Gemini 3.1 Pro Preview — Upgraded core intelligence with 2× reasoning improvement over 3 Pro, 77.1% ARC-AGI-2

Gemini 3 Series (November–December 2025)

• Gemini 3 Pro — Most intelligent stable model with state-of-the-art reasoning and vibe coding
• Gemini 3 Flash — Pro-grade reasoning at Flash-level speed, default model in Gemini app
• Gemini 3 Deep Think — Enhanced reasoning mode using parallel hypothesis exploration

Gemini 2.5 Series (March 2025)

• Gemini 2.5 Pro — Thinking model with 1M context window
• Gemini 2.5 Flash — Best price-performance for high-volume tasks
• Gemini 2.5 Flash-Lite — Most cost-efficient 2.5 model for classification/translation at scale

Gemma (Open Source)

Lightweight open models built from Gemini research, designed to run on single GPUs/TPUs.

• Gemma 3 — 1B/4B/12B/27B multimodal models with 128K context, 140+ languages (March 2025)
• Gemma 3n — Optimized for on-device execution on phones, laptops, tablets
• CodeGemma — 2B/7B models for code completion and generation
• MedGemma — Gemma 3 variant for medical text and image comprehension
• ShieldGemma 2 — 4B safety classifier for detecting harmful content

Image Models

Imagen (Image Generation)

Google's text-to-image generation family.

• Imagen 4 — Latest generation with improved detail rendering and typography (May 2025)
• Imagen 3 — High-quality photorealistic images with diverse art styles
• Nano Banana / Nano Banana Pro — Native image generation and editing within Gemini

Tools

• ImageFX — Consumer image generation tool
• Whisk — Remix subjects, scenes, and styles using reference images

Video Models

Veo (Video Generation)

Google's video generation model family.

• Veo 3.1: 8-second 720p/1080p videos with native audio, video extension, and reference images
• Veo 3: Video with synchronized audio including dialogue and ambient sounds (May 2025)
• Veo 2: Improved realism and cinematography understanding

Tools

• VideoFX: Consumer video generation tool
• Flow: AI filmmaking tool combining Veo, Imagen, and Gemini for consistent characters and scenes

Audio & Music Models

Lyria (Music Generation)

Google's music generation family from DeepMind.

• Lyria 2: High-fidelity 48kHz stereo instrumental music from text prompts (up to 30 seconds)
• Lyria RealTime: Interactive real-time music generation with live control of key, BPM, density
• Music AI Sandbox: Collaborative tools for artists to experiment with AI-assisted creation

Speech

• Speech-to-Text API: Automatic speech recognition and real-time transcription
• Text-to-Speech API: Natural-sounding speech synthesis

Embedding Models

• gemini-embedding-001: Latest Gemini embedding model, top of MTEB Multilingual leaderboard (July 2025)
• text-embedding-005: Based on Gecko research, supports dynamic embedding sizes
• text-multilingual-embedding-002: Multilingual text embeddings

Applications & Products

NotebookLM

AI-powered research and content organization tool, now built on Gemini 3 (with 3.1 Pro available for Pro/Ultra users).

• Upload PDFs, videos, websites, Google Docs as sources
• Generate Audio Overviews (podcast-style discussions), Video Overviews, Mind Maps
• Create Slide Decks, Reports, Flashcards, Quizzes, Infographics
• Data Tables with export to Google Sheets
• Deep Research queries for comprehensive analysis

NotebookLM

Google Labs

Experimental AI tools and demos.

• CC: AI productivity agent for Gmail with daily email briefings
• Disco: GenTabs for remixing browser tabs into custom apps
• Doppl: Virtual try-on for exploring personal style
• Pomelli: AI-powered scalable marketing content creation
• Opal: Build and share AI mini-apps with natural language
• Mixboard: AI-powered concepting board for ideas
• GenType: AI-generated custom alphabets
• Learn Your Way: Transform content into personalized learning experiences

Google Labs

Developer Platforms

• Google AI Studio: Free access to Gemini API for prototyping
• Vertex AI: Enterprise ML platform with 200+ models
• Google Antigravity: Agentic development platform (November 2025)
• Gemini CLI: Command-line interface for Gemini

Links

• Google AI
• Google DeepMind
• Google Labs
• Labs Experiments
• Vertex AI Documentation
• Cloud AI APIs