Sber releases GigaChat 3.5 Reasoning, first Russian open-weights reasoning model
Sber
Sber launched GigaChat 3.5 Reasoning, its first model with a full reasoning mode, built on online RL applied after SFT across six separate domain experts (math, code, agents and others) with individual reward models merged back into one model. The weights are open, and the model is available to end users and developers via the GigaChat app and API.
Why it matters
Positioned as the first openly available Russian reasoning model, a direct answer to DeepSeek R1 and OpenAI o4-mini, with a multi-expert online RL pipeline that is a technical first for a Russian lab.
Importance: 3/5
Russian flagship reasoning release with open weights; official Sber post + wide media coverage