Gemini 3.1 Flash-Lite is rolling out in preview via the Gemini API in Google AI Studio ⚡️ Our fastest and most cost-efficient Gemini 3 series model now comes with dynamic thinking to scale across tasks of any complexity: https://goo.gle/4lpiwQV 3.1 Flash-Lite is built for high-volume dev workloads at scale and outperforms 2.5 Flash in quality and speed — at 2.5X faster Time to First Answer Token and an 45% increase in output speed at the lower price of $0.25/1M input tokens and $1.50/1M output tokens. Explore the model's feature of control over cost and flexibility with thinking levels in AI Studio and start building today.
Google AI for Developers
Technology, Information and Internet
AI for every developer. So what will you build?
About us
Our goal is to equip developers with the most advanced models to build new applications, helpful tools to write better and faster code, and make it easy to integrate across platforms and devices.
- Website
-
https://goo.gle/ai-devs
External link for Google AI for Developers
- Industry
- Technology, Information and Internet
- Company size
- 10,001+ employees
Updates
-
Learn how to use Gemini Enterprise, Antigravity, and Google Cloud tools to build and scale an end-to-end agentic platform in this webinar featuring Head of AI Customer Engineering, Damian Danchenko and AI Customer Engineer, Startups Lia Passaglia ⬇️
-
Try out Hooks in Gemini CLI🪝 Take full control and customize the agentic loop to your specific needs without ever having to touch the source code. Hooks allow you to add context, validate actions, loop the agent to continue to iterate, and a lot more! Learn how to tailor Gemini CLI to your workflow: https://goo.gle/3MN0laS
-
Nano Banana 2 (aka Gemini 3.1 Flash Image) is our SoTA model that offers image generation at faster ⚡️ speeds and a lower cost with new and improved capabilities: https://goo.gle/4bgGKIr Start building today in Google AI Studio: https://goo.gle/4r2aIFz
-
See how Gemini 3 Flash and Vision-Language-Action (VLA) models work together to teach a robot how to play the board game First Orchard. This project shows how to use Gemini as the “brain” for tracking game rules and state, while the VLA handles complex physical manipulations like picking and placing fruit 🍎 Check out this detailed walkthrough from 🦝 Paul Ruiz of the data collection process, VLA training steps, and the Python game loop in action → https://goo.gle/4b6W778
-
CastFox transforms how users engage with their favorite podcasts using Gemma 3n to build features for contextual chat, smart highlights, and semantic search🎙️ By running Gemma 3n with Ollama on lightweight GPU and CPU infrastructure, the team scales their processing for English, Japanese, and Korean audio without the high cost or complexity of model retraining. Instead of fine-tuning, the developers use prompt engineering and JSON schema validation to maintain high accuracy and structured outputs. During preprocessing, episodes are transcribed and parsed into JSON-based summaries and auto-segments, creating a searchable and interactive experience for users. Practical performance benchmarks from their workflow: ◾Audio Processing: A 30-second clip is processed in ~40 seconds. ◾Summarization: 300-400 character summaries are generated in ~6 seconds. ◾Q&A Generation: Recommended questions from long-form text take ~12 seconds. Learn more about their implementation: https://goo.gle/4rydXpj