Gemini

Gemini is Google's family of AI models and the consumer assistant built on top of them, available on the web, in the mobile app, and woven into Workspace apps like Gmail, Docs and Sheets. It handles conversation, long-document analysis, image and video generation, coding, and live voice interaction.

As of mid-2026 the default model is Gemini 3.5 Flash, which reached general availability on May 19, 2026 with strong agentic and coding performance. The heavier reasoning and coding work is handled by Gemini 3.1 Pro, now the flagship model. Google also shipped Gemini Omni for native video generation and editing (May 2026) and Gemini Spark, an autonomous agent (June 2026). Note that the Gemini 3.5 Pro general release slipped to July 2026.

Capabilities span text and code generation, native image and video creation, Deep Research for multi-source reports, and Gemini Live for real-time spoken conversation. The models are natively multimodal and can work over very long contexts, which suits large codebases and long documents.

Pricing is freemium. The free tier costs nothing and includes Gemini 3.5 Flash, limited daily access to Gemini 3.1 Pro, image and video generation, Deep Research and Gemini Live. Google AI Plus is $7.99/mo, Google AI Pro is $19.99/mo (the former Gemini Advanced / Google AI Premium, with 2TB storage and higher limits), and Google AI Ultra now comes in two tiers: a $100/mo entry tier launched May 19, 2026 and a $200/mo top tier (cut from $250 at I/O 2026). The Gemini API is billed per token, roughly $2 per million input tokens and $12 per million output tokens for Gemini 3-class models.

Limitations are worth noting: free-tier access to the strongest model is capped daily, the naming across models and subscription tiers is genuinely confusing, and the top-end features sit behind expensive Ultra plans. Availability of some newer features can also vary by region.

Gemini is a good fit for people already inside Google's ecosystem, for anyone who wants image and video generation plus chat in one place, and for developers who want to build on a pay-per-token multimodal API.