Google for Developers

The next era of Android Bench!

Here is what's packed into the latest update:

8 New Models Added: Expanded the leaderboard with top industry models, including OpenAI's GPT 5.5, Anthropic's Claude Fable 5 and Sonnet 5, GLM 5.2, and Qwen 3.7 Max - with Claude Fable 5 currently leading the pack at 84.5%.

Upgraded Evaluation Methodology: To continue providing state-of-the-art evaluations of the latest model capabilities, we are standardizing the Harbor framework. This makes it easier for you to run and evaluate the benchmark yourself- providing even greater transparency.

Community Contributions: Starting today, the Android developer community can help shape the benchmark by submitting development tasks and sharing evaluations.

Check out the updated leaderboard and the full methodology breakdown in the latest blog post: goo.gle/4wB7Vqk

20 hours ago | [YT] | 227

Google for Developers

Today, we’re unlocking a whole new era for generative media workflows in the Gemini API.

We are thrilled to announce a major double-model launch, live right now in Google AI Studio and the Gemini API. Both models are built to help developers build complete multimedia pipelines:

🍌 Nano Banana 2 Lite (GA): Our fastest Gemini Image model to date. Engineered specifically for high-velocity developer pipelines and rapid drafting, it delivers text-to-image outputs in just 4 seconds at $0.034 per 1K resolution image, maintaining character consistency and pristine in-image text rendering.

🎬 Gemini Omni Flash (Public Preview): For the first time, we are bringing native multimodal video generation and conversational editing to developers. Using the Interactions API, the model keeps session context alive, letting you execute up to 3 sequential edits - like scene changes, style transfers, and product swaps - using natural language.

🔗 Chain them together: You can create seamless, multi-turn visual workflows. Use Nano Banana 2 Lite to instantly generate a series of high-speed reference assets, then pass them straight into Gemini Omni Flash to animate and refine them in motion.

The playgrounds and developer documentation are officially live in Google AI Studio. We can't wait to see what you build!

1 week ago | [YT] | 575

Google for Developers

New Kotlin challenge: a permissions bug where giving one user a new role somehow changes another user too.

Can you find out why?→ youtube.com/shorts/wq8X9YrmooU

Watch the clip and share your answer in the comments.

Skip the Carousel

1 2

1 week ago | [YT] | 337

Google for Developers

Welcome back to Google Developer News!

For this month's episode, we join host Anaya Mehta as she breaks down the biggest developer drops from Google I/O.

Highlights include:
🌐 Near real-time translation with Gemini 3.5 Live Translate
💻 Local agentic workflows with Gemma 4 12b
🍏 Gemini natively in Xcode for Apple devs

Watch the full episode here → https://youtu.be/UlbokBsjMRY

1 week ago | [YT] | 484

Google for Developers

A small CI pipeline puzzle from a very real build-system pattern.

The app moves from React 18 to React 19. The next CI build runs faster than expected. Then production crashes.

The clue is in the cache setup.→ youtube.com/shorts/qnl8-PBJNu4

Watch the clip and share the most likely cause in the comments.

Skip the Carousel

1 2

2 weeks ago | [YT] | 479

Google for Developers

A package import can look simple on the surface.

Same application code. Same package name. Different behavior after deployment.→ youtube.com/shorts/eaUd3iYkj_w

Watch the clip and share which file gets loaded in production.

Skip the Carousel

1 2

3 weeks ago | [YT] | 531

Google for Developers

What will you build this weekend?

Gemini 3.5 Live Translate is now available in Preview via the Gemini Live API, delivering low-latency speech-to-speech translation across 70+ languages in a single, continuous session.

Swipe through to see 5 ways you can use the new model.

Head straight to Google AI Studio to: goo.gle/443sU8P

Skip the Carousel

1 2 3 4 5 6 7

3 weeks ago | [YT] | 1,244

Google for Developers

Introducing DiffusionGemma, an experimental open 26B Mixture of Experts model that moves beyond traditional sequential generation to process and generate entire blocks of text simultaneously.

DiffusionGemma unlocks new value for developers:

🏃‍♀️ Generates 1,000+ tokens/sec on an NVIDIA H100 and 700+ tokens/sec on an RTX 5090
🔄 Optimizes non-linear workflows like code infilling, inline editing, and real-time self-correction
📦 Comfortably within 18GB VRAM limits of high-end dedicated consumer GPUs when quantized.
🔧 Supports native integration for MLX, vLLM, Hugging Face, and Unsloth with advanced NVIDIA NVFP4 kernel optimization

Download the Apache 2.0 weights on Hugging Face: goo.gle/4uCwBgQ
Explore the developer guide on the blog: goo.gle/3RX6zHn

4 weeks ago | [YT] | 1,381

Google for Developers

Chrome DevTools for agents changes how AI optimizes your web apps.

Instead of letting a coding agent guess from static source code and burning tokens, this allows your AI to run actual performance traces and agentic web Lighthouse audits in a real browser → goo.gle/4wVbLvc

4 weeks ago | [YT] | 33

Google for Developers

A small package management puzzle from a familiar workflow. → youtube.com/shorts/ZGP1Fzc2Q_o

Local development and CI both install the same dependency, but they end up with different versions. package.json has not changed.

The clue is in how the install is being resolved.

Watch the clip and share which file explains the mismatch.

1 month ago | [YT] | 396