FitLocally by GPUwerk

Phi · 14B

Can you run Phi-3 Medium 14B locally?

Phi-3 Medium 14B is a dense model with 14B parameters. At 4-bit or better it runs on 96 of the 120 hardware configurations listed here; the smallest is the RTX 5080 (16 GB). On a 128 GB DGX Spark it runs at Q4_K_M, at roughly 14 tokens per second (our est.).

Data as of 2 November 2026

Fastest machines for Phi-3 Medium 14B

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~90 tok/s our est.Q4_K_M · 12.3 of 31 GB
  2. 22x RTX 5090Graphics card · 64 GB~90 tok/s our est.Q4_K_M · 12.3 of 62 GB
  3. 3RTX PRO 6000 BlackwellGraphics card · 96 GB~90 tok/s our est.Q4_K_M · 12.3 of 95 GB
  4. 42x RTX PRO 6000 BlackwellGraphics card · 192 GB~90 tok/s our est.Q4_K_M · 12.3 of 190 GB
  5. 5RTX 4090Graphics card · 24 GB~50 tok/s our est.Q4_K_M · 12.3 of 23 GB
  6. 6RTX 3090 TiGraphics card · 24 GB~50 tok/s our est.Q4_K_M · 12.3 of 23 GB
  7. 72x RTX 4090Graphics card · 48 GB~50 tok/s our est.Q4_K_M · 12.3 of 46 GB
  8. 82x RTX 3090 TiGraphics card · 48 GB~50 tok/s our est.Q4_K_M · 12.3 of 46 GB
  9. 9RTX 5080Graphics card · 16 GB~48 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  10. 10RX 7900 XTXGraphics card · 24 GB~48 tok/s our est.Q4_K_M · 12.3 of 23 GB
  11. 112x RX 7900 XTXGraphics card · 48 GB~48 tok/s our est.Q4_K_M · 12.3 of 46 GB
  12. 12RTX 3090Graphics card · 24 GB~47 tok/s our est.Q4_K_M · 12.3 of 23 GB
  13. 132x RTX 3090Graphics card · 48 GB~47 tok/s our est.Q4_K_M · 12.3 of 46 GB
  14. 14RTX 5070 TiGraphics card · 16 GB~45 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  15. 15M3 UltraMac · 96–512 GB · 5 variants~41 tok/s our est.Q4_K_M · 12.3 GB
  16. 16RX 7900 XTGraphics card · 20 GB~40 tok/s our est.Q4_K_M · 12.3 of 19.0 GB
  17. 17M2 UltraMac · 64–192 GB · 7 variants~40 tok/s our est.Q4_K_M · 12.3 GB
  18. 18M1 UltraMac · 64–128 GB · 4 variants~40 tok/s our est.Q4_K_M · 12.3 GB
  19. 19RTX 4080 SUPERGraphics card · 16 GB~37 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  20. 20Tesla P100Graphics card · 16 GB~37 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  21. 21RTX 4080Graphics card · 16 GB~36 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  22. 22RTX 4070 Ti SUPERGraphics card · 16 GB~34 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  23. 23RX 9070 XTGraphics card · 16 GB~32 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  24. 24M4 MaxMac · 48–128 GB · 3 variants~27 tok/s our est.Q4_K_M · 12.3 GB
  25. 25RTX 5060 TiGraphics card · 16 GB~22 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  26. 26M4 Max · 36 GB · 32cMac · 36 GB~21 tok/s our est.Q4_K_M · 12.3 of 24 GB
  27. 27M2 MaxMac · 32–96 GB · 5 variants~20 tok/s our est.Q4_K_M · 12.3 GB
  28. 28M1 MaxMac · 32–64 GB · 4 variants~20 tok/s our est.Q4_K_M · 12.3 GB
  29. 29M3 MaxMac · 48–128 GB · 3 variants~20 tok/s our est.Q4_K_M · 12.3 GB
  30. 30Tesla T4Graphics card · 16 GB~16 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  31. 31RX 9060 XTGraphics card · 16 GB~16 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  32. 32M3 MaxMac · 36–96 GB · 2 variants~15 tok/s our est.Q4_K_M · 12.3 GB
  33. 33RTX 4060 TiGraphics card · 16 GB~14 tok/s our est.Q4_K_M · 12.3 of 15.0 GB
  34. 34M4 ProMac · 24–64 GB · 6 variants~14 tok/s our est.Q4_K_M · 12.3 GB
  35. 35DGX Spark (GB10)AI mini PC · 128 GB~14 tok/s our est.Q4_K_M · 12.3 of 122 GB
  36. 36Ryzen AI Max+ 395AI mini PC · 64–128 GB · 2 variants~13 tok/s our est.Q4_K_M · 12.3 GB
  37. 37M2 Pro · 32 GBMac · 32 GB~10 tok/s our est.Q4_K_M · 12.3 of 21 GB
  38. 38M1 Pro · 32 GBMac · 32 GB~10 tok/s our est.Q4_K_M · 12.3 of 21 GB
  39. 39M5Mac · 24–32 GB · 4 variants~7.7 tok/s slow our est.Q4_K_M · 12.3 GB
  40. 40M3 Pro · 36 GBMac · 36 GB~7.5 tok/s slow our est.Q4_K_M · 12.3 of 24 GB
  41. 41M4Mac · 24–32 GB · 2 variants~6.0 tok/s slow our est.Q4_K_M · 12.3 GB
  42. 42M3 · 24 GBMac · 24 GB~5.0 tok/s slow our est.Q4_K_M · 12.3 of 16.0 GB
  43. 43M2 · 24 GBMac · 24 GB~5.0 tok/s slow our est.Q4_K_M · 12.3 of 16.0 GB
  44. 44M5 ProMac · 24–64 GB · 6 variantsno estimateQ4_K_M · 12.3 GB
  45. 45M6Mac · 24–32 GB · 3 variantsno estimateQ4_K_M · 12.3 GB
  46. 46M5 MaxMac · 36–128 GB · 4 variantsno estimateQ4_K_M · 12.3 GB
  47. 47M5 UltraMac · 96–512 GB · 5 variantsno estimateQ4_K_M · 12.3 GB
Too big for 24 machines

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. Phi-3 Medium 14B needs about 12.3 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
14B
Active per token
14B
Type
Dense
Good for
Coding, Quality

Builds and how to run them

QuantizationLoadMin. RAMCommand
Q4_K_M11.0 GB16 GBollama run phi3:14b-medium-4k-instruct-q4_K_M

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 Phi-3 Medium 14B 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 Pro · 24 GB · 20c.

How much memory does Phi-3 Medium 14B need?

About 12.3 GB at the smallest build in our data, including a margin for the runtime and an 8K-token context. Longer contexts need more.

Does Phi-3 Medium 14B run on a DGX Spark?

Yes, at Q4_K_M. The Spark has 128 GB of unified memory, about 121.6 GB usable.