Google AI for eCommerce. Products: Gemini app: Google's consumer AI assistant with deep Workspace integration (Gmail, Docs, Sheets, Drive). Offers multiple tiers from free to Ultra with different model and tool access.. Best for: Teams already in Google Workspace — quick copy drafts in Docs, data analysis in Sheets, email triage in Gmail, and ad-hoc research with grounded web search.. Access: Gemini 3 Pro/Flash; Deep Think (mode) for some tiers; creative tools vary. Gemini API (AI Studio): Google's developer platform for building and testing apps with Gemini models. Supports text, image, audio, and video generation through a unified API with generous free tiers.. Best for: Prototyping multimodal eCommerce features — product image editing, video ad generation, catalogue enrichment from photos, and any workflow that mixes text with images or video.. Access: gemini-3-pro-preview; gemini-3-flash-preview; gemini-3-pro-image-preview; veo-3.1-generate-preview. Vertex AI: Google Cloud's enterprise AI platform. Runs the same Gemini models with added governance, fine-tuning, data residency, VPC controls, and integration with BigQuery and other GCP services.. Best for: Enterprise eCommerce teams on GCP who need compliance controls, private data pipelines, model fine-tuning, and production-grade SLAs for AI workloads.. Access: Gemini family; availability varies by region and account.. Antigravity: Google's agentic development platform for building stateful, tool-using AI agents. Not a model itself — it orchestrates Gemini models with persistent memory and action capabilities.. Best for: Building multi-step eCommerce agents that need to remember context, call external tools (inventory APIs, CRMs), and execute complex workflows autonomously.. Access: Uses Gemini models (not a model itself).. Models: Gemini 3 Pro (gemini-3-pro-preview): General eCommerce assistant, analysis, multimodal tasks (images/docs). Type: Multimodal (flagship). Strong all-rounder if your team lives in Google Workspace/Cloud. Gemini 3 Flash (gemini-3-flash-preview): Bulk copy, classification, support routing, high-volume automation. Type: Multimodal (fast/efficient). Use for speed + cost efficiency; escalate tricky cases to a deeper model. Gemini 3 Deep Think ((mode / limited access)): Hard research and complex strategy work. Type: Extended reasoning mode. Often a mode rather than a separate model ID; access can be restricted. Nano Banana Pro (gemini-3-pro-image-preview): eCommerce creative where text-on-image matters (banners, promos). Type: Image generation/editing. Nickname used in docs/community; model ID is what matters in the API. Veo 3.1 (veo-3.1-generate-preview): Short video ads, product clips, concept creative. Type: Video generation. Designed for video generation; often easier to run inside Google's creative tools.