Marcus SchulerGoogle released Gemma 4 under Apache 2.0, abandoning the custom license terms that kept enterprise teams cautious for two years. Performance appears competitive with Alibaba's Qwen 3.5 on benchmarks, but the standard licensing may prove more significant than the technical specs. Four model sizes target everything from phones to workstations.
https://www.implicator.ai/google-releases-gemma-4-under-apache-2-0-dropping-its-custom-ai-license/
