Загрузка…

Сообщить об ошибке

Gemini Cookbook: Official Quickstarts — AI Academy

AI Academy/Courses/Gemini Cookbook: Official Quickstarts

RU Sign in Sign up

IntermediateИнженерияgemini

Gemini Cookbook: Official Quickstarts

Google's official Gemini quickstarts: function calling, audio, File API, caching, code execution, embeddings, and Live API. Real Python code from the Google team.

8modules

8lessons

160 mintotal time

Developers learning Gemini API hands-onaudience

Progress

0%0 / 8 lessons

Sign in to save your progress.

Module 1

Getting Started with Google GenAI SDK

Install the new Google GenAI SDK, initialize a client, send text and multimodal prompts, set system instructions, count tokens and configure generation parameters.

First Steps with the Gemini API

Install google-genai SDK, create a client, make your first generate_content call, explore system instructions, count_tokens, temperature parameters, and multimodal image input.

Module 2

Function Calling

Connect Python functions as model tools, explore automatic and manual calling modes, inspect chat history with FunctionCall/FunctionResponse, and understand FunctionDeclaration schemas.

Function Calling in the Gemini API

Pass functions as tools, enable automatic_function_calling in ChatSession, inspect FunctionCall/FunctionResponse history, and implement a manual calling loop.

Module 3

Working with Audio

Upload audio files via the File API, work with inline audio, request transcripts with timestamps, and analyze YouTube videos.

Audio in the Gemini API

Upload MP3 via File API, pass audio inline via types.Part.from_bytes, get transcripts with timestamps, and analyze YouTube videos through file_uri.

Module 4

File API

Upload and manage files via the Gemini File API: images, text, code, GCS objects, multi-file batch uploads, and direct HTTPS URLs.

File API: Uploading and Using Files

Upload files via client.files.upload, retrieve metadata, use them in generate_content, register GCS objects, and pass direct HTTPS URLs via types.Part.from_uri.

Module 5

Context Caching

Cache large documents for repeated use: create CachedContent, manage TTL, use the cache in generate_content and chat, monitor token counts.

Context Caching: Cache Large Documents

Upload the Apollo 11 transcript, create a CachedContent with system instruction, manage TTL, use the cache in generate_content and multi-turn chat, read usage_metadata.

Module 6

Code Execution

Enable the Python code execution tool, work with files via File I/O, use code execution in chat and multimodal scenarios.

Code Execution: Python Inside Gemini

Enable ToolCodeExecution, parse responses with executable_code and code_execution_result parts, upload CSV via File API for data analysis, and use code execution in chat and with images.

Module 7

Embeddings

Generate text and multimodal embeddings, control dimensionality, compute semantic similarity, and build a simple RAG system with task_type.

Embeddings: Vector Representations in Gemini

Call embed_content for text, images, and audio; control output_dimensionality; compute cosine similarity between sentences; build an FAQ search system with task_type=RETRIEVAL_DOCUMENT.

Module 8

Live API

Real-time voice and multimodal sessions over WebSocket: text-to-text, text-to-audio, async AudioLoop, session management, and stream handling.

Live API: Real-Time Multimodal Sessions

Connect via client.aio.live.connect, implement text-to-text and text-to-audio sessions, build an AudioLoop with async send/recv, control response_modalities, and save audio to WAV.

AAI AcademyEN Start learningCatalog

Overview

AI Academy Progress Pricing

Learning

Tracks8 Courses46

Libraries

Prompts30 Workflows12 Tools22 Compare

Updated · May 14, 2026