cookbook

This repository provides a structured learning path for the Gemini API through step-by-step quickstarts, feature-specific notebooks, and practical example applications. A Google account and an API key from Google AI Studio are required to follow the guides, and many notebooks can be opened directly in Google Colab. The quickstart section begins with authentication, basic prompting, and multimodal input, then expands into individual API capabilities. These include the multimodal Live API for low-latency voice and video interaction, live extended thinking, live audio-stream translation, and Gemini Transcribe for audio files and streams with word-level timestamps, speaker diarization, custom vocabulary, and smart transcription. Media-focused guides cover Omni Flash and Veo for video generation and editing, Imagen and Nano-Banana image generation, Lyria music generation, Lyria RealTime, and text-to-speech. Other documented features include grounding with Google Search, Google Maps, YouTube, and URLs; hosted File Search for grounding responses in user data; Python code execution; the Agents API and managed Antigravity agent; Batch API for non-time-sensitive workloads; webhooks for asynchronous operation notifications; and Priority and Flex inference tiers. The repository also includes a guide for the Gemini Robotics-ER model for spatial understanding and reasoning. The examples directory combines multiple API features or third-party tools into use cases such as browser automation, book illustration, animated story generation, live API plotting and mapping, 3D spatial understanding, API cost estimation and health monitoring, and deploying a Live API instance with Gradio. End-to-end demos are linked in separate repositories, including Gemini CLI, a Flask API quickstart, a Multimodal Live API web console, a full-stack LangGraph quickstart, and Google AI Studio starter applets. The Gemini API can be called as a REST API with curl or through official SDKs for Python, Go, Node.js, Java, and C#. Enterprise Vertex AI examples are maintained separately, and contributions are documented in CONTRIBUTING.md.