Gemini Family

Gemini is Google DeepMind's flagship multimodal AI model family, introduced in December 2023 (originally as the successor to the PaLM series and Bard). Gemini models are designed as natively multimodal from the ground up - processing text, image, audio, video, and code within a single architecture. Known size variants: Ultra (strongest), Pro (balanced), Flash (fast/cheap), Nano (on-device), Flash-Lite (cheapest). Historical generations: Gemini 1.0 (December 2023), Gemini 1.5 with 1M/2M context window (February 2024, long-context breakthrough), Gemini 2.0 (December 2024, native image generation, agentic tools), Gemini 2.5 (March 2025, thinking mode), Gemini 3 (2025, reasoning), Gemini 3.1 Pro / Flash-Lite / Deep Think (2026), Gemini 3.5 Flash (2026). Specialised variants: Gemini Robotics 1.5 and Robotics-ER 1.6 (Vision-Language-Action for robotics), Gemini Robotics On-Device (edge inference for robots). The model is available in the chatbot at gemini.google.com, Google Workspace (Gemini in Docs/Gmail/Sheets), Vertex AI (for enterprise), Google AI Studio (for developers), Gemini API, and Gemini CLI. Gemini Spark is the consumer agentic variant. Competition: ChatGPT/GPT-5 (OpenAI), Claude (Anthropic), Qwen (Alibaba), DeepSeek.