2× RTX 5090
149tok/sEveryday
1 request, prose prompt, no speculative decoding Source
Intelligenceno independent value
- Experimental
- License unclear
Local AI · Recipe · 2× RTX 3090
Nemotron-3-Puzzle 75.3B with vLLM on 2× RTX 3090
Nemotron-3-Puzzle 75.3B with vLLM on 2× RTX 3090: 45.8 tok/s according to github.com (dataset as of Nov 3, 2026).
by noonghunna
Intelligence (original model) · Artificial Analysisno independent value
Engine
- Engine
- vLLM
- Quantization
- Community W4A16 compressed-tensors (danielrmay), int8_per_token_head KV
- Model family
- Nemotron-3-Puzzle
- Context
- 262,144
- Parameters
- 75.3B MoE / 9.3B aktiv (Mamba2-Hybrid)
- Creator
- noonghunna
- GitHub stars
- 2,471
- Repo updated
- Nov 2, 2026
Measurements
Every sourced value of this recipe, each with its condition and source. Bars relative to the largest value in the group.
Decode · 1 request
What you need
- Hardware
- 2 × NVIDIA RTX 3090
- Weights
- Main weightsdanielrmay/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-W4A16 other
- Engine
- vLLM
- Context
- 262,144 tokens
Notes
What matters before you rebuild it.
- Experimental
- License unclear
Sources
Related recipes
4× RTX 3090
104tok/sEveryday
1 request, prose prompt, with MTP (per recipe) Source
Intelligenceno independent value
- Experimental
- License unclear
2× RTX 3090
199tok/sEveryday
1 request, prose prompt, with DSpark (per recipe) Source
Intelligenceno independent value
2× RTX 3090
187tok/sEveryday
1 request, prose prompt, prompt 450 tokens, no speculative decoding Source
Intelligenceno independent value
2× RTX 3090
179tok/sEveryday
1 request, prose prompt, no speculative decoding Source
Intelligence18with thinkingno thinking 15
- License unclear
2× RTX 3090
151tok/sEveryday
1 request, prose prompt, no speculative decoding Source
Intelligenceno independent value
Local AI in your company?
In a workshop we work out which models and which hardware fit your tasks, and build the first agent on your infrastructure.