FitLocally by GPUwerk

Qwen · 9B

Can you run Qwen3.5 9B Instruct locally?

Qwen3.5 9B Instruct is a dense model with 9B parameters. At 4-bit or better it runs on 112 of the 120 hardware configurations listed here; the smallest is the RTX 5070 (12 GB). On a 128 GB DGX Spark it runs at Q8_0, at roughly 14 tokens per second (our est.).

Data as of 2 November 2026

Fastest machines for Qwen3.5 9B Instruct

Every machine that runs it, fastest first. Filter by type; your machine is marked. Compare machines running it side by side

  1. 1RTX 5090Graphics card · 32 GB~153 tok/s ModelFit est.Q8_0 · 12.0 of 31 GB
  2. 2RTX PRO 6000 BlackwellGraphics card · 96 GB~153 tok/s ModelFit est.Q8_0 · 12.0 of 95 GB
  3. 32x RTX 5090Graphics card · 64 GB~146 tok/s ModelFit est.Q8_0 · 12.0 of 62 GB
  4. 42x RTX PRO 6000 BlackwellGraphics card · 192 GB~146 tok/s ModelFit est.Q8_0 · 12.0 of 190 GB
  5. 5RTX 5070Graphics card · 12 GB~104 tok/s ModelFit est.Q4_K_M · 8.2 of 11.0 GB
  6. 6M5 UltraMac · 96–512 GB · 5 variants~96 tok/s ModelFit est.Q8_0 · 12.0 GB
  7. 7RTX 4090Graphics card · 24 GB~92 tok/s ModelFit est.Q8_0 · 12.0 of 23 GB
  8. 8RTX 5080Graphics card · 16 GB~91 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  9. 92x RTX 4090Graphics card · 48 GB~89 tok/s ModelFit est.Q8_0 · 12.0 of 46 GB
  10. 10RTX 3090 TiGraphics card · 24 GB~86 tok/s ModelFit est.Q8_0 · 12.0 of 23 GB
  11. 11RTX 5070 TiGraphics card · 16 GB~86 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  12. 122x RTX 3090 TiGraphics card · 48 GB~84 tok/s ModelFit est.Q8_0 · 12.0 of 46 GB
  13. 13RTX 3090Graphics card · 24 GB~81 tok/s ModelFit est.Q8_0 · 12.0 of 23 GB
  14. 142x RTX 3090Graphics card · 48 GB~79 tok/s ModelFit est.Q8_0 · 12.0 of 46 GB
  15. 15RTX 4070 TiGraphics card · 12 GB~78 tok/s ModelFit est.Q4_K_M · 8.2 of 11.0 GB
  16. 16RTX 4070 SUPERGraphics card · 12 GB~78 tok/s ModelFit est.Q4_K_M · 8.2 of 11.0 GB
  17. 17RTX 4070Graphics card · 12 GB~78 tok/s ModelFit est.Q4_K_M · 8.2 of 11.0 GB
  18. 18RX 7900 XTXGraphics card · 24 GB~75 tok/s ModelFit est.Q8_0 · 12.0 of 23 GB
  19. 192x RX 7900 XTXGraphics card · 48 GB~73 tok/s ModelFit est.Q8_0 · 12.0 of 46 GB
  20. 20RTX 4080 SUPERGraphics card · 16 GB~69 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  21. 21RTX 4080Graphics card · 16 GB~68 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  22. 22M3 UltraMac · 96–512 GB · 5 variants~66 tok/s ModelFit est.Q8_0 · 12.0 GB
  23. 23RX 7900 XTGraphics card · 20 GB~65 tok/s ModelFit est.Q8_0 · 12.0 of 19.0 GB
  24. 24RTX 4070 Ti SUPERGraphics card · 16 GB~64 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  25. 25M2 UltraMac · 64–192 GB · 7 variants~59 tok/s ModelFit est.Q8_0 · 12.0 GB
  26. 26M5 MaxMac · 48–128 GB · 3 variants~56 tok/s ModelFit est.Q8_0 · 12.0 GB
  27. 27RX 9070 XTGraphics card · 16 GB~56 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  28. 28M1 UltraMac · 64–128 GB · 4 variants~53 tok/s ModelFit est.Q8_0 · 12.0 GB
  29. 29RTX 3060Graphics card · 12 GB~53 tok/s ModelFit est.Q4_K_M · 8.2 of 11.0 GB
  30. 30M4 MaxMac · 48–128 GB · 3 variants~47 tok/s ModelFit est.Q8_0 · 12.0 GB
  31. 31RTX 5060 TiGraphics card · 16 GB~45 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  32. 32M5 Max · 36 GB · 32cMac · 36 GB~43 tok/s ModelFit est.Q8_0 · 12.0 of 24 GB
  33. 33M3 MaxMac · 48–128 GB · 3 variants~36 tok/s ModelFit est.Q8_0 · 12.0 GB
  34. 34M4 Max · 36 GB · 32cMac · 36 GB~36 tok/s ModelFit est.Q8_0 · 12.0 of 24 GB
  35. 35M2 MaxMac · 32–96 GB · 5 variants~34 tok/s ModelFit est.Q8_0 · 12.0 GB
  36. 36M2 Pro · 16 GBMac · 16 GB~32 tok/s ModelFit est.Q4_K_M · 8.2 of 10.7 GB
  37. 37M5 ProMac · 24–64 GB · 6 variants~31 tok/s ModelFit est.Q8_0 · 12.0 GB
  38. 38M1 MaxMac · 32–64 GB · 4 variants~31 tok/s ModelFit est.Q8_0 · 12.0 GB
  39. 39RX 9060 XTGraphics card · 16 GB~31 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  40. 40M1 Pro · 16 GBMac · 16 GB~28 tok/s ModelFit est.Q4_K_M · 8.2 of 10.7 GB
  41. 41RTX 4060 TiGraphics card · 16 GB~28 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  42. 42Tesla P100Graphics card · 16 GB~28 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  43. 43M3 MaxMac · 36–96 GB · 2 variants~28 tok/s ModelFit est.Q8_0 · 12.0 GB
  44. 44M6Mac · 16 GB · 2 variants~28 tok/s ModelFit est.Q4_K_M · 8.2 GB
  45. 45M4 ProMac · 24–64 GB · 6 variants~26 tok/s ModelFit est.Q8_0 · 12.0 GB
  46. 46M5Mac · 16 GB · 2 variants~26 tok/s ModelFit est.Q4_K_M · 8.2 GB
  47. 47Tesla T4Graphics card · 16 GB~24 tok/s ModelFit est.Q8_0 · 12.0 of 15.0 GB
  48. 48Ryzen AI Max+ 395 · 128 GBAI mini PC · 128 GB~22 tok/s ModelFit est.Q8_0 · 12.0 of 96 GB
  49. 49M4 · 16 GBMac · 16 GB~19 tok/s ModelFit est.Q4_K_M · 8.2 of 10.7 GB
  50. 50M2 Pro · 32 GBMac · 32 GB~19 tok/s ModelFit est.Q8_0 · 12.0 of 21 GB
  51. 51M6Mac · 24–32 GB · 3 variants~19 tok/s ModelFit est.Q8_0 · 12.0 GB
  52. 52M1 Pro · 32 GBMac · 32 GB~17 tok/s ModelFit est.Q8_0 · 12.0 of 21 GB
  53. 53M3 · 16 GBMac · 16 GB~17 tok/s ModelFit est.Q4_K_M · 8.2 of 10.7 GB
  54. 54M2Mac · 16–24 GB · 2 variants~16 tok/s ModelFit est.Q4_K_M · 8.2 GB
  55. 55M5Mac · 24–32 GB · 4 variants~16 tok/s ModelFit est.Q8_0 · 12.0 GB
  56. 56M3 ProMac · 18–36 GB · 2 variants~15 tok/s ModelFit est.Q8_0 · 12.0 GB
  57. 57DGX Spark (GB10)AI mini PC · 128 GB~14 tok/s our est.Q8_0 · 12.0 of 122 GB
  58. 58Ryzen AI Max+ 395 · 64 GBAI mini PC · 64 GB~13 tok/s our est.Q8_0 · 12.0 of 48 GB
  59. 59M4Mac · 24–32 GB · 2 variants~12 tok/s ModelFit est.Q8_0 · 12.0 GB
  60. 60M3 · 24 GBMac · 24 GB~10 tok/s ModelFit est.Q8_0 · 12.0 of 16.0 GB
  61. 61M1 · 16 GBMac · 16 GB~9.9 tok/s slow ModelFit est.Q4_K_M · 8.2 of 10.7 GB
Too big for 8 machines

RTX 5060, RTX 4060, RTX 3070, RTX 3060 Ti, RTX 2070, M3 · 8 GB, M2 · 8 GB, M1 · 8 GB. Qwen3.5 9B Instruct needs about 8.2 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
9B
Active per token
9B
Type
Dense
Good for
Quality, Coding, Reasoning

Builds and how to run them

QuantizationLoadMin. RAMCommand
Q4_K_M7.0 GB12 GBollama run qwen3.5:9b
Q8_010.7 GB16 GBollama run qwen3.5:9b-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 Qwen3.5 9B Instruct 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 M2 Pro · 16 GB.

How much memory does Qwen3.5 9B Instruct need?

About 8.2 GB at the smallest build in our data, including a margin for the runtime and an 8K-token context. Longer contexts need more.

Does Qwen3.5 9B Instruct run on a DGX Spark?

Yes, at Q8_0. The Spark has 128 GB of unified memory, about 121.6 GB usable.