← All models

Nemotron Cascade 8B Thinking

NVIDIA · 2025-12-08 · 8.2B parameters

Nemotron Cascade 8B Thinking is NVIDIA's open-weight reasoning-only model post-trained from Qwen3-8B-Base with SFT and sequential, domain-wise reinforcement learning (Cascade RL), released on December 8, 2025. The BF16 checkpoint contains 8,190,735,360 parameters in the Qwen3 dense architecture (36 layers, GQA) with a 32,768-token native context that NVIDIA extends to 64K with YaRN for evaluation.

Benchmark scores