Google has published Gemma 4 as an open-weight model and made it commercially available under the Apache 2.0 license according to its official blog post. The model is hosted on Hugging Face at https://huggingface.co/collections/google/gemma-4 and Ollama at https://ollama.com/library/gemma4. Primary source coverage states only these availability details and the license type.

Google previously released Gemma 2 in June 2024 under a more restricted license that limited certain commercial uses (https://blog.google/technology/ai/google-gemma-2/). Meta released Llama 3.1 in July 2024 under a license permitting commercial deployment (https://ai.meta.com/blog/meta-llama-3-1/). Coverage of the Gemma 4 announcement did not reference these parallel releases or the evolution in Google's licensing approach from its first Gemma model.

The Apache 2.0 license allows modification, distribution and commercial application with minimal restrictions, a fact stated directly in the source document. No performance benchmarks or architecture details beyond the open-weight designation appear in the primary announcement.