Google for Developers
@googledevs
Discover the latest developer tools, resources, events, and announcements to help you build smarter, ship faster. ๐
313 Following 2.6M Followers
A major update to Tunix for scaling Agentic RL is here ๐ง
The new asynchronous, decoupled rollout engine solves multi-turn bottlenecks on JAX/TPU with generic environment support and built-in microsecond-level RL profiling.
Deep dive โ
Show more
Meet LiteRT.js: @Googleโs new Edge AI runtime for the web! โจ
We've made it easier to convert from PyTorch to #WebAI# using the common LiteRT model format. Powered by #WebGPU#, WebAssembly, & #WebNN#! Time to upgrade from TensorFlow.js and build the future.
Read more โก๏ธ
Show more
AlphaEvolve is now Generally Available (GA) via @GoogleCloud.
Built in partnership with @GoogleDeepMind, this Gemini-powered agent brings autonomous algorithm optimization to your hardest engineering and research problems.
Check out the frontier science and tech impact so far:
๐ฅ๏ธ GPU kernels: Generating exascale GPU kernels on the Frontier supercomputer
โ๏ธ Quantum computing: Discovering brand-new quantum error-correction schemes
๐งช Drug discovery: Accelerating molecular simulations 4x for faster drug discovery
๐ Forecasting: Improving forecasting accuracy while slashing runtimes by 90%
Read our launch post to start optimizing your workflows today:
Show more
Build production-ready, multi-agent applications with @golang ๐ค
The Agent Development Kit for Go 2.0 runs single agents and complex graphs on the same execution model.
โ
Dynamic orchestration written in plain Go
โ
Native human-in-the-loop primitives
โ
Built-in retry policies
โ
Unified telemetry spans
Learn more:
Show more
Build agents that can see, reason, and take action across browser, desktop and mobile environments. Gemini 3.5 Flash now features built-in computer use capabilities:
๐ฑ Custom client-side functions for human-in-the-loop takeover
๐ Configurable action-level safety policies
๐ก๏ธ Intent arguments to maximize transparency
Learn more in the blog โ
Show more
Build cross-language multi-agent AI pipelines using Google ADK and the Agent-to-Agent (A2A) protocol.
Python for AI extraction. Go for strict policy validation.
Open-source Contract Compliance multi-agent engine:
Show more
The best collaborators don't just respond. They anticipate.
As coding agents become more capable, evaluating them requires more than measuring whether they complete a task. It means understanding how they help developers move faster, stay in flow, and make progress on meaningful work.
Learn how @julesagent is measuring what matters in the latest blog:
Show more
The Interactions API is officially generally available (GA), taking center stage as the primary interface for building with Gemini models and agents.
This release introduces key new capabilities built to scale developer workflows:
โก๏ธ Managed agents for seamless orchestration
โก๏ธ Background execution for log-running asynchronous tasks
โก๏ธ Native support of Gemini Omni coming up in the API
๐ฅ Get started with the docs:
๐ Read the blog: ๐ปโจ
Show more
Building agentic apps is easy. Making them production-ready is hard. ๐ ๏ธ
Announcing Genkit Middleware: A composable way to intercept, extend, and harden your agentic apps.
โ
Retries & Fallbacks
โ
Human-in-the-loop approvals
โ
Full observability
Now live for TypeScript, Go, and Dart!
Show more
Gemma 4: Now up to 3x Faster. โก
Same quality, way more speed. Our new MTP drafters allow Gemma 4 to predict multiple tokens at once, effectively tripling your output speed without compromising intelligence.
Show more
