Modelofficial release
Parameters12B
Context256K
Input → outputtext + image + audio → text + image + audio
Licenseapache-2.0
PrecisionBF16
ArchitectureGemma4UnifiedForConditionalGeneration
ReleasedMay 23, 2026
Downloads75.7K

Other sizes and formats in the family

ModelParamsPrecisionDownloads
google/gemma-4-12B-it12BBF161.7M

Compare side by side →

Built on this model

ModelParamsPrecisionDownloads
google/gemma-4-12B-it12BBF161.7M
jialinyyzz/humanizer12BBF16103.7K

This release in the news

No coverage of this exact release yet.

Recent news about google/gemma-4-12B

  1. Oct 15, 2026 · Opinion / analysis
    QUASAR 4-bit QAT checkpoints for Qwen3.5, Gemma 4 and Muse Glimmer: closer to BF16 than Google’s and NVIDIA’s 4-bit releases
  2. Oct 15, 2026 · Research paper
    Prism-1: Position- and Rival-Independent Scoring for System One Models
  3. Oct 12, 2026 · Research paper
    Releasing the Inner Jev: A Simple Recipe for Open Decision Models
  4. Oct 7, 2026 · Research paper
    Real Long-Term Memory for AI: A 50-Million-Token Window That Is Faster and Cheaper Than Recompute