GigaChat Lite (GigaChat-20B-A3B)

Training compute

9.9×10²² FLOP

Parameters

20B

Published

Dec 13, 2024

GigaChat Lite (GigaChat-20B-A3B) is an AI model developed by Sber (Russia), first published in December 2024. It works in the language domain, on tasks such as language modeling/generation, question answering and chat.

Training it took an estimated 9.9×10²² FLOP of compute. The model has 20,000,000,000 parameters.

Access: Open weights (unrestricted). Its weights are openly available. Epoch AI rates the confidence of this record as speculative.

Full record

Organization: Sber
Country of organization: Russia
Domain: Language
Task: Language modeling/generation, Question answering, Chat
Training compute: 9.9×10²² FLOP
Parameters: 20,000,000,000
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Epoch confidence: Speculative