Google Releases Gemma 4 Open Models
Google released Gemma 4 on April 2, 2026, a family of four open-weight models spanning phones to data centers and licensed under Apache 2.0. The lineup includes E2B and E4B for edge, a 26B Mixture-of-Experts for speed, and a 31B Dense that ranks #3 on Arena AI's text leaderboard, with weights available on Hugging Face and Google AI Studio. The release strengthens U.S. open-source AI competition and enables local, commercial deployments.
Key Points
- 1Releases four Gemma 4 models (E2B, E4B, 26B MoE, 31B Dense) under Apache 2.0
- 2Positions U.S. open-source AI as stronger contender versus Chinese models like Qwen and DeepSeek
- 3Enables practitioners to run, modify, and commercialize high-quality models locally with large contexts
Scoring Rationale
Official, same-day release from Google/DeepMind with Apache 2.0 licensing and usable weights; high novelty, scope, actionability, credibility, and relevance. Slightly tempered because proprietary systems still lead hardest benchmarks, but the announcement meaningfully shifts open-model dynamics.
Sources
Public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems
