Apa yang dijawab halaman ini
Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at...
- Google · google/gemma-4-26b-a4b-it:free
- text+image+video->text · rute model global
- 262.144 context · US$0,00 input