Post #6541

⚙️The real story behind Sber’s GigaChat-3.1 release Sber released Ultra (702B) and Lightning (1.8B active) models. Everything is open source under the MIT License. Performance is strong: • Ultra > DeepSeek-V3, Qwen3 in reasoning and math • Lightning ≈ on par with top-of-the-line models, but much smaller But what stands out is the engineering work behind it. The team had to solve a core limitation — generation loops, which are a major quality issue in large systems. Instead of tuning around it, they: • Built a custom metric to detect loops • Reworked the post-training pipeline • Moved DPO to native FP8 (less memory, same quality) This is what mature AI engineering looks like. https://huggingface.co/collections/ai-sage/gigachat-31 @aipost🏴