Google launched Gemma 4 under Apache 2.0, positioning it for reasoning, agentic workflows, multimodality, and on-device use. François Chollet called it Google's strongest open model yet, and day-0 ecosystem support landed across vLLM, llama.cpp, and Ollama simultaneously. The license shift to a "real" open-weights release drew significant community attention.
Models
[AINews] Good Friday
Google shipped Gemma 4 under Apache 2.0 with day-0 adoption across vLLM, llama.cpp, and Ollama—a genuine open-source play for multimodal reasoning and agentic workflows that François Chollet called the company's strongest open model.
Friday, April 3, 2026 12:00 PM UTC2 MIN READSOURCE: Latent.SpaceBY sys://pipeline
Tags
models
/// RELATED
Infrastructure4d ago
Our agent found a bug with WireGuard in Google Kubernetes Engine
Lovable's AI agents uncovered a concurrency bug in Google's WireGuard integration for GKE by analyzing logs at scale, revealing a concurrent map-access panic that caused random pod crashes and required an MTU mismatch fix.
War4d ago
Fast16 Malware
Fast16, a newly uncovered pre-Stuxnet US state-sponsored malware, sabotaged Iranian computational research by silently corrupting high-precision physics simulations—revealing early-stage sophistication in cyber-warfare infrastructure targeting critical academic and research infrastructure.