AvitoTech/avibe

AvitoTech/avibe is an 8 billion parameter large language model developed by AvitoTech, based on the Qwen3-8B-Base architecture. Optimized for Russian language tasks and the Avito domain, it features a custom tokenizer for improved tokenization density and faster processing of Russian text. The model demonstrates superior performance over the instruct version of Qwen3-8B on numerous Russian benchmarks, excelling in mathematical problem-solving and Function Calling capabilities.