Llama-3.1 Nemotron Ultra 253B vs Qwen3.5-397B-A17B

Llama-3.1 Nemotron Ultra 253B

NVIDIA

Llama-3.1-Nemotron-Ultra-253B-v1 is a 253-billion-parameter model from NVIDIA, derived from Meta's Llama 3.1 405B using neural architecture search (NAS) compression combined with NVIDIA's Nemotron post-training pipeline, which recovers and exceeds the base model's capability after structural compression. Released in April 2025, it supports toggling between a standard instruction mode and an extended reasoning mode via system prompt, allowing the same model to handle both rapid responses and deliberate chain-of-thought tasks. It is the flagship of the Nemotron family, available open-weight on HuggingFace and through NVIDIA NIM for enterprise inference.

Alibaba / Qwen

Qwen3.5-397B-A17B is a 397-billion-parameter mixture-of-experts model from Alibaba's Qwen team, released in February 2026 as the open-weight flagship of the Qwen3.5 series, featuring 17 billion active parameters per forward pass through a hybrid linear-attention and sparse-MoE architecture based on Gated Delta Networks. The model was co-trained on text, images, and video using early fusion, making it natively multimodal across a 262K token context window, while achieving significantly higher inference throughput than comparable dense models due to its sparse computation design. At release it was one of the most capable open-weight models publicly available, offered under Apache 2.0 and accessible through Alibaba's DashScope API as the Qwen3.5-Plus endpoint.