Building Multimodal Apps with Gemini 2 0 and Gemini SDK
summary
Live demos show Gemini as a multimodal, large-context LLM that “thinks” before answering and ships with controls like thinking budget and a model optimizer; examples include image/video/audio understanding (counting ~44 people, reading laundry tags, remembering door code 1170), streaming text and code generation via the Gemini SDK and AI Studio/Collab calling Vertex AI. Models named: Gemini 2.5 Pro (2M token context) plus Flash, Imagine 3/V2 for images/video and Lydia for music. Most demos work, though a few stall from network hiccups.
AI-written summary · about
chapters (18)
- 0:00Introduction and demo
- 4:44Gemini 2.5 overview
- 9:25Live coding demo
- 12:39AI Studio live demo
- 17:49Building with live API
- 20:49Hands-on labs setup
- 23:00Google Cloud credits and project
- 27:26Google Cloud project creation
- 31:14Google Colab setup
- 33:43Lab walkthrough and execution
- 37:27Gemini SDK integration
- 39:56Multimodal streaming demo
- 41:54Vertex AI setup
- 44:41System prompts and safety filters
- 46:49Structured output with Pydantic
- 50:46Advanced multimodality features
- 52:30AI Studio vs Vertex AI vs Firebase Studio
- 54:35Summary and next steps
related talks
54mGemini, Google's Large Language Model
51mGemini, Google's Large Language Model
01h 16mI'm in my Gemini Era!
01h 45mAI Engineering with the Google Gemini 2.5 Model Family
18mIntroduction to Gemini APIs and AI Studio
18mLet's test Gemini, Google's AI in an Android application!
22mMilliseconds to Magic: Real‑Time Workflows using the Gemini Live API and Pipecat
21mAdd Generative AI capabilities to your Android app