Qwen · 32B
Can you run Qwen3 32B locally?
Qwen3 32B is a dense model with 32B parameters. At 4-bit or better it runs on 63 of the 120 hardware configurations listed here; the smallest is the RTX 4090 (24 GB). On a 128 GB DGX Spark it runs at Q4_K_M, at roughly 7.4 tokens per second (our est.).
Data as of 2 November 2026
Fastest machines for Qwen3 32B
Every machine that runs it, fastest first. Filter by type; your machine is marked. Compare machines running it side by side
- 1RTX 5090Graphics card · 32 GB~71 tok/s ModelFit est.Q4_K_M · 22 of 31 GB
- 2RTX PRO 6000 BlackwellGraphics card · 96 GB~44 tok/s ModelFit est.Q8_0 · 37 of 95 GB
- 32x RTX 5090Graphics card · 64 GB~43 tok/s ModelFit est.Q8_0 · 37 of 62 GB
- 42x RTX PRO 6000 BlackwellGraphics card · 192 GB~43 tok/s ModelFit est.Q8_0 · 37 of 190 GB
- 5RTX 4090Graphics card · 24 GB~40 tok/s ModelFit est.Q4_K_M · 22 of 23 GB
- 6RTX 3090 TiGraphics card · 24 GB~38 tok/s ModelFit est.Q4_K_M · 22 of 23 GB
- 7RTX 3090Graphics card · 24 GB~36 tok/s ModelFit est.Q4_K_M · 22 of 23 GB
- 8RX 7900 XTXGraphics card · 24 GB~35 tok/s ModelFit est.Q4_K_M · 22 of 23 GB
- 9M5 UltraMac · 96–512 GB · 5 variants~27 tok/s ModelFit est.Q8_0 · 37 GB
- 102x RTX 4090Graphics card · 48 GB~24 tok/s ModelFit est.Q8_0 · 37 of 46 GB
- 11M5 Max · 48 GB · 40cMac · 48 GB~24 tok/s ModelFit est.Q4_K_M · 22 of 36 GB
- 122x RTX 3090 TiGraphics card · 48 GB~23 tok/s ModelFit est.Q8_0 · 37 of 46 GB
- 132x RTX 3090Graphics card · 48 GB~21 tok/s ModelFit est.Q8_0 · 37 of 46 GB
- 142x RX 7900 XTXGraphics card · 48 GB~21 tok/s ModelFit est.Q8_0 · 37 of 46 GB
- 15M4 Max · 48 GB · 40cMac · 48 GB~20 tok/s ModelFit est.Q4_K_M · 22 of 36 GB
- 16M5 Max · 36 GB · 32cMac · 36 GB~18 tok/s ModelFit est.Q4_K_M · 22 of 24 GB
- 17M3 UltraMac · 96–512 GB · 5 variants~18 tok/s ModelFit est.Q8_0 · 37 GB
- 18M2 UltraMac · 64–192 GB · 7 variants~17 tok/s ModelFit est.Q8_0 · 37 GB
- 19M3 MaxMac · 48–128 GB · 3 variants~15 tok/s ModelFit est.Q4_K_M · 22 GB
- 20M4 Max · 36 GB · 32cMac · 36 GB~15 tok/s ModelFit est.Q4_K_M · 22 of 24 GB
- 21M1 UltraMac · 64–128 GB · 4 variants~15 tok/s ModelFit est.Q8_0 · 37 GB
- 22M2 MaxMac · 64–96 GB · 3 variants~15 tok/s ModelFit est.Q4_K_M · 22 GB
- 23M5 MaxMac · 64–128 GB · 2 variants~14 tok/s ModelFit est.Q8_0 · 37 GB
- 24M1 MaxMac · 64 GB · 2 variants~13 tok/s ModelFit est.Q4_K_M · 22 GB
- 25M5 ProMac · 48–64 GB · 4 variants~13 tok/s ModelFit est.Q4_K_M · 22 GB
- 26M4 MaxMac · 64–128 GB · 2 variants~12 tok/s ModelFit est.Q8_0 · 37 GB
- 27M3 MaxMac · 36–96 GB · 2 variants~12 tok/s ModelFit est.Q4_K_M · 22 GB
- 28M4 ProMac · 48–64 GB · 4 variants~11 tok/s ModelFit est.Q4_K_M · 22 GB
- 29Ryzen AI Max+ 395 · 128 GBAI mini PC · 128 GB~9.2 tok/s slow ModelFit est.Q4_K_M · 22 of 96 GB
- 30DGX Spark (GB10)AI mini PC · 128 GB~7.4 tok/s slow our est.Q4_K_M · 22 of 122 GB
- 31Ryzen AI Max+ 395 · 64 GBAI mini PC · 64 GB~7.0 tok/s slow our est.Q4_K_M · 22 of 48 GB
- 32M3 Pro · 36 GBMac · 36 GB~6.3 tok/s slow ModelFit est.Q4_K_M · 22 of 24 GB
Too big for 57 machines
M2 Max · 32 GB · 38c, M2 Max · 32 GB · 30c, M1 Max · 32 GB · 32c, M1 Max · 32 GB · 24c, M2 Pro · 32 GB, M1 Pro · 32 GB, M6 · 32 GB · 12c, M5 · 32 GB · 10c, M5 · 32 GB · 8c, M4 · 32 GB, RX 7900 XT, M5 Pro · 24 GB · 20c, M5 Pro · 24 GB · 16c, M4 Pro · 24 GB · 20c, M4 Pro · 24 GB · 16c, M6 · 24 GB · 12c, M6 · 24 GB · 10c, M5 · 24 GB · 10c, M5 · 24 GB · 8c, M4 · 24 GB, M3 · 24 GB, M2 · 24 GB, RTX 5080, RTX 5070 Ti, RTX 5060 Ti, RTX 4080 SUPER, RTX 4080, RTX 4070 Ti SUPER, RTX 4060 Ti, Tesla T4, Tesla P100, RX 9070 XT, RX 9060 XT, M3 Pro · 18 GB, RTX 5070, RTX 4070 Ti, RTX 4070 SUPER, RTX 4070, RTX 3060, M2 Pro · 16 GB, M1 Pro · 16 GB, M6 · 16 GB · 12c, M6 · 16 GB · 10c, M5 · 16 GB · 10c, M5 · 16 GB · 8c, M4 · 16 GB, M3 · 16 GB, M2 · 16 GB, M1 · 16 GB, RTX 5060, RTX 4060, RTX 3070, RTX 3060 Ti, RTX 2070, M3 · 8 GB, M2 · 8 GB, M1 · 8 GB. Qwen3 32B needs about 22 GB at its smallest build.
Speeds are single-user decode estimates, not measurements. "ModelFit est." comes from ModelFit's model fitted on published runs (CC BY 4.0); "our est." is memory bandwidth divided by bytes read per token, times 0.55. Memory is what the best build needs, against what the machine can give a model. Method
About the model
- Parameters
- 32B
- Active per token
- 32B
- Type
- Dense
- Good for
- Quality, Coding
Builds and how to run them
| Quantization | Load | Min. RAM | Command |
|---|---|---|---|
| Q4_K_M | 20 GB | 32 GB | ollama run qwen3:32b-q4_K_M |
| Q8_0 | 35 GB | 64 GB | ollama run qwen3:32b-q8_0 |
Load, minimum RAM and Ollama commands: ModelFit, CC BY 4.0. Load is memory with the model loaded, before long contexts.
Questions
Can I run Qwen3 32B on a Mac?
Yes on a Mac with enough memory: the smallest Apple configuration listed here that runs it at 4-bit or better is the M5 Max · 36 GB · 32c.
How much memory does Qwen3 32B need?
About 22 GB at the smallest build in our data, including a margin for the runtime and an 8K-token context. Longer contexts need more.
Does Qwen3 32B run on a DGX Spark?
Yes, at Q4_K_M. The Spark has 128 GB of unified memory, about 121.6 GB usable.