Google Research unveils AI tools for science, health, and edge devices
At Research@ 2026, Google showcased AI systems that amplify human ingenuity across planetary intelligence, scientific discovery, and wearables.
AI & ML
Frontier models, research results, agents, and developer tooling.
Hugging Face and Cerebras partner on Gemma 4 for voice AI
The collaboration aims to use Cerebras's inference hardware with Hugging Face's platform to enable real-time voice AI on Gemma 4.
ScarfBench: A New Benchmark for AI-Driven Java Framework Migration
IBM Research introduces a standardized benchmark to evaluate how well AI agents handle migrating enterprise Java applications between frameworks.
LeRobot v0.6.0 Adds Tools to Imagine, Evaluate, and Improve Robots
The new release of Hugging Face's robotics framework centers on simulation, evaluation, and iterative refinement for robot policies.
Import AI Highlights Fable's GPU Kernel Writing and Analog Compute
The latest Import AI edition points to AI-written GPU kernels, broader automation, and analog computing as signals of a shifting compute landscape.
Hugging Face Announces Major Kernel Updates
A new blog post from Hugging Face details major updates to its kernels, which are essential for efficient AI model execution.
DeepMind and A24 Launch First-of-Its-Kind AI-Film Research Partnership
The collaboration marks the first formal research tie-up between a major AI lab and a prestigious film studio.
AI Gateways Ranked for 2026: Bifrost, Portkey, LiteLLM and More
A new ranking compares nine AI gateways on latency, pricing, licensing, governance, and deployment to help teams choose an infrastructure layer for LLM access.
Liquid AI's 2B-Active MoE Model Targets On-Device Agents
Liquid AI's new open-weight model packs 24B parameters but activates only 2B, aiming to run capable reasoning agents locally on devices like Apple's M5 Max.
NVIDIA's UK Ecosystem: $250M Compute, Earthshot, and AI Deployments
NVIDIA's UK ecosystem moves from ambition to application with a $250M compute commitment, Earthshot partnership, and a wave of enterprise and startup AI deployments.
NVIDIA-Powered AI Agents Help European Cities Run Smarter Operations
European cities are deploying NVIDIA-based AI agents and digital twins to improve urban planning, energy use, and public services.
Agentic AI Adoption Rising in EU Enterprises, OpenAI Reports
OpenAI's analysis finds agentic AI moving into real use across EU businesses, with untapped potential for deeper value.
MUFG Deploys ChatGPT Enterprise to Build an AI-Native Financial Group
The move signals a major institutional bet on OpenAI's enterprise technology for both internal workflow improvements and new customer-facing services.
AI-Designed Enzymes Can Be Tuned to Cut Specific Proteins
New protein-design tools rebuilt the catalytic cores of two natural proteases, producing custom enzymes that cleave chosen targets.
Australian Payments Plus Uses OpenAI Tools to Ease Payments Complexity
A case study from OpenAI describes how AP+ applies ChatGPT Enterprise and Codex to save time and improve quality while keeping humans in the loop.
How OpenAI, Anthropic, and Two Sigma Approach Financial Agent Safety
At AI Engineer New York, frontier labs and financial firms converged on a shared playbook for giving agents autonomy without losing human oversight.
Hugging Face Adds One-Click Launch into SageMaker Studio
A new integration lets Hugging Face users jump straight into Amazon SageMaker Studio, reducing the manual steps between model development and AWS deployment.
Hugging Face Adds Native vLLM Backend to Transformers
The integration promises Transformers users a native-speed path to vLLM's optimized inference.