Gemma 4 hits Apache 2.0 with agentic power

Gemma 4 is a category-defining open-weights release that finally brings Google's best reasoning tech to the local-first community.

• –Apache 2.0 license makes it a viable, friction-free alternative to Llama for commercial use
• –Specifically optimized for agents with native function calling and structured output
• –31B model rivals much larger proprietary systems in coding and math benchmarks
• –Massive 256K context window enables complex RAG and long-running autonomous tasks
• –Day-one support across Ollama, vLLM, and Apple Silicon ensures immediate local utility