BREAKING
Gov't Users Shift to NVIDIA Nemotron
Proprietary vs Open-Weight
Proprietary
Old
●
OpenAI, Anthropic
●
Data leaves the org
●
Charged for tokens
Open-Weight
New
●
NVIDIA Nemotron
●
Keep the weights
●
Fine-tune on own data
Palantir + NVIDIA Sovereign Engine
1
Palantir AIP + Foundry
↓
2
NVIDIA AI stack
↓
3
Nemotron weights
↓
4
Air-gapped deploy
Nemotron 3 Model Sizes
Nano
30
Super
110
Ultra
525
0
%
AIME 2025 w/ tools
0
%
LiveCodeBench v6
0
M
Context tokens
Open Weights for Sovereign AI
AI NEWS BLITZ
Palantir's CEO says some U.S. government customers are moving to NVIDIA's open-weight Nemotron models.