# Google DeepMind Releases Gemma 4, Open-Source AI Models That Rival Industry Giants Under Apache 2.0 License

*2026-04-03 21:01 UTC · Technology · 1 min read*

> Google DeepMind released four open-weight Gemma 4 AI models under Apache 2.0, with the 31B model ranking #3 globally among open models.

Google DeepMind released Gemma 4 on April 2, a family of four open-weight AI models built from the same research behind Gemini 3. The 31B Dense model ranks #3 among all open models on the Arena AI text leaderboard, while the 26B Mixture of Experts model holds the #6 spot — making them the most capable open-source models Google has ever released.

The four models (E2B, E4B, 26B MoE, and 31B Dense) natively process video, images, and text across 140+ languages with advanced reasoning and multi-step planning capabilities. Edge models support 128K context windows while larger models extend to 256K tokens. All are released under the permissive Apache 2.0 license and available on Google AI Studio, Hugging Face, Kaggle, and Ollama.

## Sources

- [blog.google](https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/)
- [dataconomy.com](https://dataconomy.com/2026/04/03/google-releases-open-source-ai-model-gemma-4-for-developers/)

---
TechTools News · https://techtools.cz/tools/news/article/1250-google-deepmind-releases-gemma-4-open-source-ai-models-that-rival-industry
Česká verze: https://techtools.cz/tools/news/clanek/1301-google-deepmind-vydal-gemma-4-open-source-ai-modely-konkurujici-gigantum-pod
