Google Drops Gemma 4 12B (June 3) as an Apache-2.0 Mid-Sized Open-Weight Model

Artificial Analysis·medium signal

Google released Gemma 4 12B on June 3, 2026 as the new mid-sized member of the Gemma 4 family, slotting between the smaller E2B/E4B variants and the 26B/31B models. The line ships under a clean Apache 2.0 license with native multimodality and up to a 256K context window on larger variants, and the 31B reportedly scores 85.2% on MMLU Pro. For local builders, the 12B size targets the sweet spot of capability versus single-GPU deployability.