BenchMark'd
No ratings yet
Be the first to review this model
NVIDIA's high-accuracy reasoning model with approximately 100B parameters and 10B active per token. Designed for building transparent, efficient, and specialized agentic AI systems at scale. A strong choice for enterprise multi-agent deployments requiring robust reasoning capabilities.
Released
January 15, 2026
Parameters
100B (MoE, 10B active)
Context
128K
Pricing
Open Source