← Back to all models

Phi mini MoE instruct

Chat model by Microsoft · 7.65B parameters

Runs Excellently

Phi mini MoE instruct fits comfortably in your 24 GB Mac using Q8_0 quantization, using 38% of your RAM.

~111 tok/s

Q8_0 · 9.0 GB

Quantization Options

QuantizationMemorySpeedFits?
Q8_0 Recommended 9.0 GB~111 tok/s
Q6_K 7.1 GB~145 tok/s
Q5_K_M 6.2 GB~171 tok/s
Q4_K_M 5.4 GB~201 tok/s
Q3_K_M 4.7 GB~242 tok/s
Q2_K 3.8 GB~314 tok/s

Specifications

Parameters7.6B
Architecturephimoe
Context Length4K
CategoryChat
Mixture of Experts 16 experts, 2 active
Formatgguf
Minimum RAM4.3 GB
HuggingFace Downloads106,680

Which Mac Can Run Phi mini MoE instruct?

Minimum Mac

MacBook Air — M1 16GB

Q8_0 · ~28 tok/s · 56% RAM

MacQuanttok/sRAM %
MacBook Air · M1 16GB Q8_0 ~28 56%
MacBook Air · M2 16GB Q8_0 ~41 56%
MacBook Air · M3 16GB Q8_0 ~41 56%
MacBook Air · M4 16GB Q8_0 ~49 56%
MacBook Air · M5 16GB Q8_0 ~62 56%
MacBook Air · M6 16GB Q8_0 ~83 56%
MacBook Pro 14" · M1 Pro 16GB Q8_0 ~81 56%
MacBook Pro 14" · M2 Pro 16GB Q8_0 ~81 56%
MacBook Pro 14" · M5 16GB Q8_0 ~62 56%
MacBook Pro 14" · M6 16GB Q8_0 ~83 56%
MacBook Pro 16" · M1 Pro 16GB Q8_0 ~81 56%
MacBook Pro 16" · M2 Pro 16GB Q8_0 ~81 56%
Mac Mini · M1 16GB Q8_0 ~28 56%
Mac Mini · M2 16GB Q8_0 ~41 56%
Mac Mini · M2 Pro 16GB Q8_0 ~81 56%
Mac Mini · M4 16GB Q8_0 ~49 56%
Mac Mini · M6 16GB Q8_0 ~83 56%
iMac · M1 16GB Q8_0 ~28 56%
iMac · M3 16GB Q8_0 ~41 56%
iMac · M4 16GB Q8_0 ~49 56%
iMac · M6 16GB Q8_0 ~83 56%
MacBook Pro 14" · M3 Pro 18GB Q8_0 ~61 50%
MacBook Pro 16" · M3 Pro 18GB Q8_0 ~61 50%
MacBook Air · M2 24GB Q8_0 ~41 38%
MacBook Air · M3 24GB Q8_0 ~41 38%
MacBook Air · M4 24GB Q8_0 ~49 38%
MacBook Air · M5 24GB Q8_0 ~62 38%
MacBook Air · M6 24GB Q8_0 ~83 38%
MacBook Pro 14" · M4 Pro 24GB Q8_0 ~111 38%
MacBook Pro 14" · M5 24GB Q8_0 ~62 38%
MacBook Pro 14" · M5 Pro 24GB Q8_0 ~125 38%
MacBook Pro 14" · M6 24GB Q8_0 ~83 38%
MacBook Pro 14" · M6 Pro 24GB Q8_0 ~166 38%
MacBook Pro 16" · M4 Pro 24GB Q8_0 ~111 38%
MacBook Pro 16" · M5 Pro 24GB Q8_0 ~125 38%
MacBook Pro 16" · M6 Pro 24GB Q8_0 ~166 38%
Mac Mini · M2 24GB Q8_0 ~41 38%
Mac Mini · M4 24GB Q8_0 ~49 38%
Mac Mini · M4 Pro 24GB Q8_0 ~111 38%
Mac Mini · M6 24GB Q8_0 ~83 38%
Mac Mini · M6 Pro 24GB Q8_0 ~166 38%
iMac · M3 24GB Q8_0 ~41 38%
iMac · M4 24GB Q8_0 ~49 38%
iMac · M6 24GB Q8_0 ~83 38%
MacBook Air · M4 32GB Q8_0 ~49 28%
MacBook Air · M5 32GB Q8_0 ~62 28%
MacBook Air · M6 32GB Q8_0 ~83 28%
MacBook Pro 14" · M1 Pro 32GB Q8_0 ~81 28%
MacBook Pro 14" · M2 Pro 32GB Q8_0 ~81 28%
MacBook Pro 14" · M5 32GB Q8_0 ~62 28%
MacBook Pro 14" · M6 32GB Q8_0 ~83 28%
MacBook Pro 16" · M1 Pro 32GB Q8_0 ~81 28%
MacBook Pro 16" · M1 Max 32GB Q8_0 ~162 28%
MacBook Pro 16" · M2 Pro 32GB Q8_0 ~81 28%
MacBook Pro 16" · M2 Max 32GB Q8_0 ~162 28%
Mac Mini · M2 Pro 32GB Q8_0 ~81 28%
Mac Mini · M4 32GB Q8_0 ~49 28%
Mac Mini · M6 32GB Q8_0 ~83 28%
Mac Studio · M1 Max 32GB Q8_0 ~162 28%
Mac Studio · M2 Max 32GB Q8_0 ~162 28%
iMac · M4 32GB Q8_0 ~49 28%
iMac · M6 32GB Q8_0 ~83 28%
MacBook Pro 14" · M3 Pro 36GB Q8_0 ~61 25%
MacBook Pro 14" · M5 Max 36GB Q8_0 ~249 25%
MacBook Pro 16" · M3 Pro 36GB Q8_0 ~61 25%
MacBook Pro 16" · M3 Max 36GB Q8_0 ~162 25%
MacBook Pro 16" · M4 Max 36GB Q8_0 ~222 25%
MacBook Pro 16" · M5 Max 36GB Q8_0 ~249 25%
Mac Studio · M4 Max 36GB Q8_0 ~222 25%
MacBook Pro 14" · M4 Pro 48GB Q8_0 ~111 19%
MacBook Pro 14" · M5 Pro 48GB Q8_0 ~125 19%
MacBook Pro 14" · M5 Max 48GB Q8_0 ~249 19%
MacBook Pro 14" · M6 Pro 48GB Q8_0 ~166 19%
MacBook Pro 14" · M6 Max 48GB Q8_0 ~374 19%
MacBook Pro 16" · M3 Max 48GB Q8_0 ~162 19%
MacBook Pro 16" · M4 Pro 48GB Q8_0 ~111 19%
MacBook Pro 16" · M4 Max 48GB Q8_0 ~222 19%
MacBook Pro 16" · M5 Pro 48GB Q8_0 ~125 19%
MacBook Pro 16" · M5 Max 48GB Q8_0 ~249 19%
MacBook Pro 16" · M6 Pro 48GB Q8_0 ~166 19%
MacBook Pro 16" · M6 Max 48GB Q8_0 ~374 19%
Mac Mini · M4 Pro 48GB Q8_0 ~111 19%
Mac Mini · M6 Pro 48GB Q8_0 ~166 19%
Mac Studio · M6 Max 48GB Q8_0 ~374 19%
MacBook Pro 14" · M5 Pro 64GB Q8_0 ~125 14%
MacBook Pro 14" · M5 Max 64GB Q8_0 ~249 14%
MacBook Pro 14" · M6 Pro 64GB Q8_0 ~166 14%
MacBook Pro 14" · M6 Max 64GB Q8_0 ~374 14%
MacBook Pro 16" · M1 Max 64GB Q8_0 ~162 14%
MacBook Pro 16" · M2 Max 64GB Q8_0 ~162 14%
MacBook Pro 16" · M3 Max 64GB Q8_0 ~162 14%
MacBook Pro 16" · M4 Max 64GB Q8_0 ~222 14%
MacBook Pro 16" · M5 Max 64GB Q8_0 ~249 14%
MacBook Pro 16" · M6 Pro 64GB Q8_0 ~166 14%
MacBook Pro 16" · M6 Max 64GB Q8_0 ~374 14%
Mac Mini · M6 Pro 64GB Q8_0 ~166 14%
Mac Studio · M1 Max 64GB Q8_0 ~162 14%
Mac Studio · M1 Ultra 64GB Q8_0 ~325 14%
Mac Studio · M2 Max 64GB Q8_0 ~162 14%
Mac Studio · M2 Ultra 64GB Q8_0 ~325 14%
Mac Studio · M4 Max 64GB Q8_0 ~222 14%
Mac Studio · M6 Max 64GB Q8_0 ~374 14%
MacBook Pro 16" · M2 Max 96GB Q8_0 ~162 9%
MacBook Pro 16" · M3 Max 96GB Q8_0 ~162 9%
Mac Studio · M2 Max 96GB Q8_0 ~162 9%
Mac Studio · M2 Ultra 96GB Q8_0 ~325 9%
Mac Pro · M2 Ultra 96GB Q8_0 ~325 9%
MacBook Pro 14" · M5 Max 128GB Q8_0 ~249 7%
MacBook Pro 14" · M6 Max 128GB Q8_0 ~374 7%
MacBook Pro 16" · M3 Max 128GB Q8_0 ~162 7%
MacBook Pro 16" · M4 Max 128GB Q8_0 ~222 7%
MacBook Pro 16" · M5 Max 128GB Q8_0 ~249 7%
MacBook Pro 16" · M6 Max 128GB Q8_0 ~374 7%
Mac Studio · M1 Ultra 128GB Q8_0 ~325 7%
Mac Studio · M2 Ultra 128GB Q8_0 ~325 7%
Mac Studio · M4 Max 128GB Q8_0 ~222 7%
Mac Studio · M4 Ultra 128GB Q8_0 ~332 7%
Mac Studio · M6 Max 128GB Q8_0 ~374 7%
Mac Pro · M2 Ultra 128GB Q8_0 ~325 7%
Mac Pro · M4 Ultra 128GB Q8_0 ~332 7%
MacBook Pro 14" · M6 Max 192GB Q8_0 ~374 5%
MacBook Pro 16" · M6 Max 192GB Q8_0 ~374 5%
MacBook Pro 16" · M6 Ultra 192GB Q8_0 ~748 5%
Mac Studio · M2 Ultra 192GB Q8_0 ~325 5%
Mac Studio · M4 Ultra 192GB Q8_0 ~332 5%
Mac Studio · M6 Max 192GB Q8_0 ~374 5%
Mac Studio · M6 Ultra 192GB Q8_0 ~748 5%
Mac Pro · M2 Ultra 192GB Q8_0 ~325 5%
Mac Pro · M4 Ultra 192GB Q8_0 ~332 5%
Mac Pro · M6 Ultra 192GB Q8_0 ~748 5%
MacBook Pro 16" · M6 Ultra 256GB Q8_0 ~748 4%
Mac Studio · M4 Ultra 256GB Q8_0 ~332 4%
Mac Studio · M6 Ultra 256GB Q8_0 ~748 4%
Mac Pro · M4 Ultra 256GB Q8_0 ~332 4%
Mac Pro · M6 Ultra 256GB Q8_0 ~748 4%
MacBook Pro 16" · M6 Ultra 384GB Q8_0 ~748 2%
Mac Studio · M6 Ultra 384GB Q8_0 ~748 2%
Mac Pro · M6 Ultra 384GB Q8_0 ~748 2%
MacBook Air · M1 8GB Q5_K_M ~43 78%
MacBook Air · M2 8GB Q5_K_M ~63 78%
MacBook Air · M3 8GB Q5_K_M ~63 78%
Mac Mini · M1 8GB Q5_K_M ~43 78%
Mac Mini · M2 8GB Q5_K_M ~63 78%
iMac · M1 8GB Q5_K_M ~43 78%
iMac · M3 8GB Q5_K_M ~63 78%

Run Phi mini MoE instruct locally on your Mac

ToolPiper downloads, manages, and runs models with one click. Apple Silicon optimized.

Get ToolPiper — Free