Gemini

Gemini is Google's flagship conversational AI assistant and the parent product of a family that includes Gemini Deep Research, Tasks in Gemini, Gemini CLI, Jules and Gemini Notebook, formerly NotebookLM. Google names its app models by plan rather than by version: the free plan runs a Flash-Lite model, Google AI Plus a Flash model, and Google AI Pro and Ultra add the Pro model and Gemini 4 Argon, Google's newest frontier model for complex tasks, which rolled out to Pro and Ultra subscribers on 13 October 2026. Argon is replacing Deep Think, Google's former Ultra-only advanced reasoning option, and Gemini Omni is Google's model for video creation.

Capabilities span conversational chat and multimodal input and output across text, image, audio and video. Google's help center lists the context window as 32k tokens without a plan, 128k tokens on Google AI Plus and 1 million tokens on Google AI Pro and Ultra. Deep Research is the built-in research agent, running autonomous multi-source report generation. Paid plans raise access to Nano Banana in Google Search, and Pro adds higher access to Gemini 4 Argon and agentic capabilities there. Creative features cover image generation with Nano Banana 2 and redoing images with Nano Banana Pro, video generation via Gemini Omni and Veo 3.1, and music generation via Lyria 3.5. Gemini Live provides voice, Canvas provides collaborative editing, and skills, which launched globally on 30 September 2026 and will replace Gems, save reusable instructions for repeated tasks. Google's connected-apps table lists more than 80 apps and services Gemini can work with, from Adobe and Airtable to Xero and Zoho, alongside Gemini inside Gmail, Docs, Sheets, Slides and Drive. Google Flow provides an AI creative studio with tiered credits. Tasks in Gemini, the renamed Gemini Spark, automates workflows in the background, including scheduled tasks. The Gemini app runs on macOS and on Windows 10 or later from 10 September 2026, and since 13 October 2026 the macOS app has Google Antigravity built in for coding and prototyping.

Gemini's Model Context Protocol support sits in the Gemini API rather than in the consumer app. Google hosts a public first-party MCP server at gemini-api-docs-mcp.dev, the Gemini Docs MCP, which a coding agent installs with npx add-mcp and which exposes a documentation search function returning live API definitions and integration patterns. Google pairs the server with a gemini-api-dev skill for coding assistants. In the other direction, the Interactions API can connect to remote MCP servers over Streamable HTTP, taking a server name and url in the tools configuration, and Google's Deep Research agent in the same API accepts remote MCP servers as a tool. Google's Interactions API documentation states that Gemini 3 does not yet support remote MCP and that support is coming, while the Deep Research agent accepts remote MCP servers as a tool.

Pricing has four consumer tiers, published in full on Google's own subscriptions page. Free with a Google account provides a Flash-Lite model, image generation and editing, Deep Research, Gemini Live, Canvas, Gems, limited Google Flow access and Gemini Notebook, with 15GB of storage. Google AI Plus at $4.99/month adds a Flash model, 2x higher usage limits than Free, video generation, Daily Brief, 200 Flow credits and 400GB of storage. Google AI Pro at $19.99/month adds Gemini 4 Argon and the Pro model, 4x higher usage limits than Free, 1,000 Flow credits, higher Jules limits, Google Antigravity access, Gemini in Chrome with auto browse and 5TB of storage. Google AI Ultra starts at $99.99/month for 5x Pro usage limits, with a $199.99/month option at 20x, the highest access to Gemini 4 Argon, cloud storage starting at 20TB, Project Genie and first access to Tasks in Gemini in select countries. Usage limits in the Gemini app are compute-based, refresh every 5 hours up to a weekly limit, and can be extended by buying AI credits. The subscriptions page lists every tier at a monthly rate, and Google One also sells some Google AI plans on annual billing.

For organizations, Google has included Gemini in its Google Workspace business plans since January 2025, when it dropped the separate Gemini Business and Gemini Enterprise add-ons. The Business tab of the same subscriptions page now sells Gemini Enterprise, a separate agent platform for companies, from $21 per seat per month. Google's SOC 2 Type II, ISO 27001 and ISO 42001 scope documents each name the Gemini app. HIPAA is the exception: Google's HIPAA Included Functionality list covers the Gemini app, excluding Gemini in Chrome, and the Gemini Mac app for Workspace customers under a Business Associate Agreement, so it does not extend to the consumer Google AI plans.

Gemini is best for people and teams who already work in Gmail, Docs and Drive and want one assistant across research, writing and creative media. Teams standardized on Word, Excel, Outlook and Teams should consider Microsoft 365 Copilot, which works inside those Microsoft apps. Several premium features are geographically limited: Google offers Ultra subscribers first access to Tasks in Gemini in select countries, and Chrome auto browse is rolling out in preview to Pro and Ultra subscribers in the U.S., so subscribers elsewhere get a narrower feature set than the plan page lists.

Current state Q4 2026: Gemini 4 Argon rolled out to Google AI Pro and Ultra on 13 October 2026 and is replacing Deep Think. Gemini Spark has been renamed Tasks in Gemini, skills launched on 30 September 2026 and will replace Gems, and the macOS app gained a built-in Antigravity coding agent. Remote MCP in the Interactions API does not yet cover Gemini 3.