Tutorials
Google just casually disrupted the open-source AI narrative…
Google released Gemma 4 on April 2, 2026 — a family of four open-weight models built on the same research as Gemini 3 and licensed under the permissive Apache 2.0 license, marking a significant shi...
Google released Gemma 4 on April 2, 2026 — a family of four open-weight models built on the same research as Gemini 3 and licensed under the permissive Apache 2.0 license, marking a significant shift from the more restrictive terms of previous Gemma versions. The lineup spans from lightweight edge variants (E2B/E4B) capable of running offline on phones and Raspberry Pi devices, up to a 31B Dense model that runs on a consumer GPU and achieves top-three rankings on global open-model leaderboards. Key technical innovations include memory-efficient quantization (TurboQuant) and per-layer embeddings that allow the models to deliver competitive intelligence at a fraction of the parameter count typical of rival open models.
Related
- Share your Gemma 4 builds or the model variants you’re training in the replies below!
- We love seeing what you’ve built with Gemma 4, the open model family that we released last week. Here are a few fun examples, described by t…
- Gemma 4 on AI Gateway
- The pace at which useful things are shipping also seems to be accelerating. Model releases are coming faster, of course, but so are signific…
Source: tutorials