~/best-gpu-for/phi-4-mini
Microsoft

Best GPU for Phi-4 Mini

Ranked recommendations from the 42-GPU catalog for running Phi-4 Mini (3.8B params). Top picks by quality, then by cost.

Top picks.

$ ./rank --by quality
Tightest budget
RTX 3070
8GB · runs at Q8_0
4.0 GB weights · 1.5 GB headroom
Best quality
RTX 3060 12GB
12GB · runs at FP16/BF16
smallest card supporting the best available quant
Apple Silicon
M2 Max 64
64GB · runs at FP16/BF16
smallest M-series that fits
Datacenter (FP16 ambition)
H100 80GB
80GB · runs at FP16/BF16
best-quality datacenter card that fits

Full ranked list.

$ ./vrambudget --rank phi-4-mini
GPUVRAMBest quantWeights at 15% safetyFit
nvidiaRTX 3060 12GB12GBFP16/BF16
7.6fitsnvidiaRTX 3080 Ti12GBFP16/BF16
7.6fitsnvidiaRTX 407012GBFP16/BF16
7.6fitsnvidiaRTX 4070 Super12GBFP16/BF16
7.6fitsnvidiaRTX 507012GBFP16/BF16
7.6fitsintelArc B58012GBFP16/BF16
7.6fitsnvidiaRTX 4060 Ti 16GB16GBFP16/BF16
7.6fitsnvidiaRTX 4070 Ti Super16GBFP16/BF16
7.6fitsnvidiaRTX 408016GBFP16/BF16
7.6fitsnvidiaRTX 4080 Super16GBFP16/BF16
7.6fitsnvidiaRTX 5070 Ti16GBFP16/BF16
7.6fitsnvidiaRTX 508016GBFP16/BF16
7.6fitsnvidiaRTX 309024GBFP16/BF16
7.6fitsnvidiaRTX 3090 Ti24GBFP16/BF16
7.6fitsnvidiaRTX 409024GBFP16/BF16
7.6fitsamdRX 7900 XTX24GBFP16/BF16
7.6fitsintelArc Pro B6024GBFP16/BF16
7.6fitsnvidiaRTX 509032GBFP16/BF16
7.6fits

Next steps.

$ ./next
$ checkCan I run Phi-4 Mini?see all quants →full quant matrix$ tuneTune the mathopen calculator →/calc?model=phi-4-mini

Discussion.

$ gh discussion list

// sign in with github to leave a comment. threads live in the repo's discussions tab.