Gemini
Also known as: Google Gemini, Gemini models
Gemini launched in December 2023 as Google's answer to GPT-4 and the successor to LaMDA and PaLM 2. From the start, it was designed to be natively multimodal, processing text, images, audio, video, and code in a single architecture rather than bolting vision onto a text model. The original lineup used Ultra/Pro/Nano naming; since Gemini 2 and 3, the main tiers are Pro (maximum capability), Flash (fast and efficient), and Flash-Lite (high-volume, cost-optimized).
The Gemini 3 generation launched in November 2025 and quickly became the dominant family for Google's products. Gemini 3.1 Pro arrived in February 2026 as a major reasoning upgrade, achieving 77.1% on ARC-AGI-2 — more than double Gemini 3 Pro's score — and 80.6% on SWE-bench Verified. In May 2026, Google I/O introduced Gemini 3.5 Flash, the newest Flash model, which delivers near-Pro intelligence at Flash-tier cost and speed and beat Gemini 3.1 Pro on 11 of 15 published coding and agentic benchmarks. The Gemini app switched to Gemini 3 Flash as its default model in late 2025.
Gemini's practical advantages over other frontier families include the industry's largest commercially reliable context window (1 million tokens), native Google Search grounding, deep integration with Google Workspace, Android, and Chrome, and competitive pricing. Flash-tier models in particular are among the most cost-effective options for high-volume workloads. The Gemini API is available through Google AI Studio (free tier for developers) and Vertex AI (enterprise), and the latest generation supports the Computer Use tool for agentic browser automation.