Local model run decision

Can Nvidia GeForce RTX 4090 (AD102) 24GB run DeepSeek R1 Distill Qwen 32B?

Not at the recommended setting. Use Q2_K at 8K context.

0.59GB more memory is required.

24.59GBCalculated requirement
24GBUsable VRAM / RAM
Q4_K_MRecommended quant
8KContext
Calculated, not benchmarked.

Related compatibility answers