Sberbank releases 432B MoE model GigaChat3.5 with day-0 GGUF
Russian tech company Sberbank released a new called GigaChat3.5-432B-A28B. It has 432 billion total parameters, but uses a (MoE) so only 28 billion parameters are actually active when generating a response. A base, pretrained-only version was released alongside it.
Notably, a GGUF version was made available on day one — GGUF is a compact file format that lets a model run on llama.cpp, a program for running models on personal computers. However, this GGUF support has not yet been merged into llama.cpp's main branch; it currently requires building from a separate pull request (PR).
Key points
- Sberbank released GigaChat3.5-432B-A28B, an MoE model with 432B total but only 28B
- A pretrained-only base version was released alongside it
- Day-0 GGUF format support was provided for local/offline use
- GGUF support isn't yet merged into llama.cpp's main branch — requires building from a pull request (PR)