← Back to all models

Phi tiny MoE instruct

Chat model by Microsoft · 3.76B parameters

Runs Excellently

Phi tiny MoE instruct fits comfortably in your 24 GB Mac using Q8_0 quantization, using 20% of your RAM.

~226 tok/s

Q8_0 · 4.7 GB

Quantization Options

QuantizationMemorySpeedFits?
Q8_0 Recommended 4.7 GB~226 tok/s
Q6_K 3.8 GB~296 tok/s
Q5_K_M 3.3 GB~348 tok/s
Q4_K_M 2.9 GB~409 tok/s
Q3_K_M 2.6 GB~494 tok/s
Q2_K 2.1 GB~640 tok/s

Specifications

Parameters3.8B
Architecturephimoe
Context Length4K
CategoryChat
Mixture of Experts 16 experts, 2 active
Formatgguf
Minimum RAM2.1 GB
HuggingFace Downloads631,814

Which Mac Can Run Phi tiny MoE instruct?

Minimum Mac

MacBook Air — M1 8GB

Q8_0 · ~56 tok/s · 59% RAM

MacQuanttok/sRAM %
MacBook Air · M1 8GB Q8_0 ~56 59%
MacBook Air · M2 8GB Q8_0 ~83 59%
MacBook Air · M3 8GB Q8_0 ~83 59%
Mac Mini · M1 8GB Q8_0 ~56 59%
Mac Mini · M2 8GB Q8_0 ~83 59%
iMac · M1 8GB Q8_0 ~56 59%
iMac · M3 8GB Q8_0 ~83 59%
MacBook Air · M1 16GB Q8_0 ~56 29%
MacBook Air · M2 16GB Q8_0 ~83 29%
MacBook Air · M3 16GB Q8_0 ~83 29%
MacBook Air · M4 16GB Q8_0 ~99 29%
MacBook Air · M5 16GB Q8_0 ~127 29%
MacBook Air · M6 16GB Q8_0 ~169 29%
MacBook Pro 14" · M1 Pro 16GB Q8_0 ~165 29%
MacBook Pro 14" · M2 Pro 16GB Q8_0 ~165 29%
MacBook Pro 14" · M5 16GB Q8_0 ~127 29%
MacBook Pro 14" · M6 16GB Q8_0 ~169 29%
MacBook Pro 16" · M1 Pro 16GB Q8_0 ~165 29%
MacBook Pro 16" · M2 Pro 16GB Q8_0 ~165 29%
Mac Mini · M1 16GB Q8_0 ~56 29%
Mac Mini · M2 16GB Q8_0 ~83 29%
Mac Mini · M2 Pro 16GB Q8_0 ~165 29%
Mac Mini · M4 16GB Q8_0 ~99 29%
Mac Mini · M6 16GB Q8_0 ~169 29%
iMac · M1 16GB Q8_0 ~56 29%
iMac · M3 16GB Q8_0 ~83 29%
iMac · M4 16GB Q8_0 ~99 29%
iMac · M6 16GB Q8_0 ~169 29%
MacBook Pro 14" · M3 Pro 18GB Q8_0 ~124 26%
MacBook Pro 16" · M3 Pro 18GB Q8_0 ~124 26%
MacBook Air · M2 24GB Q8_0 ~83 20%
MacBook Air · M3 24GB Q8_0 ~83 20%
MacBook Air · M4 24GB Q8_0 ~99 20%
MacBook Air · M5 24GB Q8_0 ~127 20%
MacBook Air · M6 24GB Q8_0 ~169 20%
MacBook Pro 14" · M4 Pro 24GB Q8_0 ~226 20%
MacBook Pro 14" · M5 24GB Q8_0 ~127 20%
MacBook Pro 14" · M5 Pro 24GB Q8_0 ~254 20%
MacBook Pro 14" · M6 24GB Q8_0 ~169 20%
MacBook Pro 14" · M6 Pro 24GB Q8_0 ~339 20%
MacBook Pro 16" · M4 Pro 24GB Q8_0 ~226 20%
MacBook Pro 16" · M5 Pro 24GB Q8_0 ~254 20%
MacBook Pro 16" · M6 Pro 24GB Q8_0 ~339 20%
Mac Mini · M2 24GB Q8_0 ~83 20%
Mac Mini · M4 24GB Q8_0 ~99 20%
Mac Mini · M4 Pro 24GB Q8_0 ~226 20%
Mac Mini · M6 24GB Q8_0 ~169 20%
Mac Mini · M6 Pro 24GB Q8_0 ~339 20%
iMac · M3 24GB Q8_0 ~83 20%
iMac · M4 24GB Q8_0 ~99 20%
iMac · M6 24GB Q8_0 ~169 20%
MacBook Air · M4 32GB Q8_0 ~99 15%
MacBook Air · M5 32GB Q8_0 ~127 15%
MacBook Air · M6 32GB Q8_0 ~169 15%
MacBook Pro 14" · M1 Pro 32GB Q8_0 ~165 15%
MacBook Pro 14" · M2 Pro 32GB Q8_0 ~165 15%
MacBook Pro 14" · M5 32GB Q8_0 ~127 15%
MacBook Pro 14" · M6 32GB Q8_0 ~169 15%
MacBook Pro 16" · M1 Pro 32GB Q8_0 ~165 15%
MacBook Pro 16" · M1 Max 32GB Q8_0 ~331 15%
MacBook Pro 16" · M2 Pro 32GB Q8_0 ~165 15%
MacBook Pro 16" · M2 Max 32GB Q8_0 ~331 15%
Mac Mini · M2 Pro 32GB Q8_0 ~165 15%
Mac Mini · M4 32GB Q8_0 ~99 15%
Mac Mini · M6 32GB Q8_0 ~169 15%
Mac Studio · M1 Max 32GB Q8_0 ~331 15%
Mac Studio · M2 Max 32GB Q8_0 ~331 15%
iMac · M4 32GB Q8_0 ~99 15%
iMac · M6 32GB Q8_0 ~169 15%
MacBook Pro 14" · M3 Pro 36GB Q8_0 ~124 13%
MacBook Pro 14" · M5 Max 36GB Q8_0 ~508 13%
MacBook Pro 16" · M3 Pro 36GB Q8_0 ~124 13%
MacBook Pro 16" · M3 Max 36GB Q8_0 ~331 13%
MacBook Pro 16" · M4 Max 36GB Q8_0 ~451 13%
MacBook Pro 16" · M5 Max 36GB Q8_0 ~508 13%
Mac Studio · M4 Max 36GB Q8_0 ~451 13%
MacBook Pro 14" · M4 Pro 48GB Q8_0 ~226 10%
MacBook Pro 14" · M5 Pro 48GB Q8_0 ~254 10%
MacBook Pro 14" · M5 Max 48GB Q8_0 ~508 10%
MacBook Pro 14" · M6 Pro 48GB Q8_0 ~339 10%
MacBook Pro 14" · M6 Max 48GB Q8_0 ~761 10%
MacBook Pro 16" · M3 Max 48GB Q8_0 ~331 10%
MacBook Pro 16" · M4 Pro 48GB Q8_0 ~226 10%
MacBook Pro 16" · M4 Max 48GB Q8_0 ~451 10%
MacBook Pro 16" · M5 Pro 48GB Q8_0 ~254 10%
MacBook Pro 16" · M5 Max 48GB Q8_0 ~508 10%
MacBook Pro 16" · M6 Pro 48GB Q8_0 ~339 10%
MacBook Pro 16" · M6 Max 48GB Q8_0 ~761 10%
Mac Mini · M4 Pro 48GB Q8_0 ~226 10%
Mac Mini · M6 Pro 48GB Q8_0 ~339 10%
Mac Studio · M6 Max 48GB Q8_0 ~761 10%
MacBook Pro 14" · M5 Pro 64GB Q8_0 ~254 7%
MacBook Pro 14" · M5 Max 64GB Q8_0 ~508 7%
MacBook Pro 14" · M6 Pro 64GB Q8_0 ~339 7%
MacBook Pro 14" · M6 Max 64GB Q8_0 ~761 7%
MacBook Pro 16" · M1 Max 64GB Q8_0 ~331 7%
MacBook Pro 16" · M2 Max 64GB Q8_0 ~331 7%
MacBook Pro 16" · M3 Max 64GB Q8_0 ~331 7%
MacBook Pro 16" · M4 Max 64GB Q8_0 ~451 7%
MacBook Pro 16" · M5 Max 64GB Q8_0 ~508 7%
MacBook Pro 16" · M6 Pro 64GB Q8_0 ~339 7%
MacBook Pro 16" · M6 Max 64GB Q8_0 ~761 7%
Mac Mini · M6 Pro 64GB Q8_0 ~339 7%
Mac Studio · M1 Max 64GB Q8_0 ~331 7%
Mac Studio · M1 Ultra 64GB Q8_0 ~661 7%
Mac Studio · M2 Max 64GB Q8_0 ~331 7%
Mac Studio · M2 Ultra 64GB Q8_0 ~661 7%
Mac Studio · M4 Max 64GB Q8_0 ~451 7%
Mac Studio · M6 Max 64GB Q8_0 ~761 7%
MacBook Pro 16" · M2 Max 96GB Q8_0 ~331 5%
MacBook Pro 16" · M3 Max 96GB Q8_0 ~331 5%
Mac Studio · M2 Max 96GB Q8_0 ~331 5%
Mac Studio · M2 Ultra 96GB Q8_0 ~661 5%
Mac Pro · M2 Ultra 96GB Q8_0 ~661 5%
MacBook Pro 14" · M5 Max 128GB Q8_0 ~508 4%
MacBook Pro 14" · M6 Max 128GB Q8_0 ~761 4%
MacBook Pro 16" · M3 Max 128GB Q8_0 ~331 4%
MacBook Pro 16" · M4 Max 128GB Q8_0 ~451 4%
MacBook Pro 16" · M5 Max 128GB Q8_0 ~508 4%
MacBook Pro 16" · M6 Max 128GB Q8_0 ~761 4%
Mac Studio · M1 Ultra 128GB Q8_0 ~661 4%
Mac Studio · M2 Ultra 128GB Q8_0 ~661 4%
Mac Studio · M4 Max 128GB Q8_0 ~451 4%
Mac Studio · M4 Ultra 128GB Q8_0 ~677 4%
Mac Studio · M6 Max 128GB Q8_0 ~761 4%
Mac Pro · M2 Ultra 128GB Q8_0 ~661 4%
Mac Pro · M4 Ultra 128GB Q8_0 ~677 4%
MacBook Pro 14" · M6 Max 192GB Q8_0 ~761 2%
MacBook Pro 16" · M6 Max 192GB Q8_0 ~761 2%
MacBook Pro 16" · M6 Ultra 192GB Q8_0 ~1,523 2%
Mac Studio · M2 Ultra 192GB Q8_0 ~661 2%
Mac Studio · M4 Ultra 192GB Q8_0 ~677 2%
Mac Studio · M6 Max 192GB Q8_0 ~761 2%
Mac Studio · M6 Ultra 192GB Q8_0 ~1,523 2%
Mac Pro · M2 Ultra 192GB Q8_0 ~661 2%
Mac Pro · M4 Ultra 192GB Q8_0 ~677 2%
Mac Pro · M6 Ultra 192GB Q8_0 ~1,523 2%
MacBook Pro 16" · M6 Ultra 256GB Q8_0 ~1,523 2%
Mac Studio · M4 Ultra 256GB Q8_0 ~677 2%
Mac Studio · M6 Ultra 256GB Q8_0 ~1,523 2%
Mac Pro · M4 Ultra 256GB Q8_0 ~677 2%
Mac Pro · M6 Ultra 256GB Q8_0 ~1,523 2%
MacBook Pro 16" · M6 Ultra 384GB Q8_0 ~1,523 1%
Mac Studio · M6 Ultra 384GB Q8_0 ~1,523 1%
Mac Pro · M6 Ultra 384GB Q8_0 ~1,523 1%

Run Phi tiny MoE instruct locally on your Mac

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

Get ToolPiper — Free