4× RTX PRO 6000
113tok/sPeak
1 request, best value reported by source Source
Intelligence29with thinking
- License unclear
Local AI · Recipe · 4× RTX PRO 6000
MiniMax-M3 428B-A23B MXFP4 with SGLang on 4× RTX PRO 6000
MiniMax-M3 428B-A23B MXFP4 with SGLang on 4× RTX PRO 6000: 39.2 tok/s according to github.com (dataset as of Nov 3, 2026).
by ambientlight
Engine
- Engine
- SGLang
- Quantization
- MXFP4 Experten + MXFP8 Linears
- Model family
- MiniMax-M3
- Context
- 1,048,576
- Parameters
- 428B MoE / 23B active
- Creator
- ambientlight
- GitHub stars
- 5
- Repo updated
- Aug 20, 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
Time to first token
Shorter is better.
What you need
- Hardware
- 4 × NVIDIA RTX PRO 6000 Blackwell
- Weights
- Main weightsolka-fi/MiniMax-M3-MXFP4 other
- Engine
- SGLang
- Context
- 1,048,576 tokens
Related recipes
4× RTX PRO 6000
87tok/sPeak
1 request, best value reported by source Source
Intelligence29with thinking
- License unclear
4× DGX Spark
38tok/sPeak
1 request, best value reported by source Source
Intelligence29with thinking
- License unclear
2× DGX Spark
36.6tok/sPeak
1 request, JSON prompt, context 196k Source
Intelligence29with thinking
- License unclear
4× DGX Spark
33.7tok/sPeak
1 request, best value reported by source Source
Intelligence29with thinking
- Experimental
- Custom kernel required
- License unclear
3× DGX Spark
10.5tok/sPeak
1 request, best value reported by source Source
Intelligence29with thinking
- License unclear
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.