AnalysisAI ModelsJune 24, 2026
Community fine-tune of Gemma4 26B released on Hugging Face
HauhauCS uploaded a fine-tuned variant of Google's Gemma4 26B MoE model (4B active parameters) to Hugging Face. The model uses QAT and is labeled 'Uncensored' with balanced MTP. It has garnered 49 likes and over 28,000 downloads.