~/model/huggingfacemicrosoft/phi-4
Microsoft

Phi-4

Microsoft's synthetic-data 14B. Punches above its weight on math and code.

Parameters
14.7B
Family
Microsoft
Context
16K tokens
FP16 weights
29GB
// where you can run it
phi4LM StudiovLLMMLXoMLX
// hugging face stats (cached daily)
744K downloads · 2K likes · license: mit · updated 5 months ago

What you need to run this.

$ ./vrambudget --model phi-4 --by quant
FP1629 GB
RTX A600048 GB · fitsRTX 6000 Ada48 GB · fitsL40S48 GB · fitsRadeon Pro W790048 GB · fits
Q8_016 GB
RTX 309024 GB · fitsRTX 3090 Ti24 GB · fitsRTX 409024 GB · fitsRX 7900 XTX24 GB · fits
Q6_K12 GB
RTX 4060 Ti 16GB16 GB · fitsRTX 4070 Ti Super16 GB · fitsRTX 408016 GB · fitsRTX 4080 Super16 GB · fits
Q5_K_M10 GB
RTX 4060 Ti 16GB16 GB · fitsRTX 4070 Ti Super16 GB · fitsRTX 408016 GB · fitsRTX 4080 Super16 GB · fits
Q4_K_M8.3 GB
RTX 3060 12GB12 GB · fitsRTX 3080 Ti12 GB · fitsRTX 407012 GB · fitsRTX 4070 Super12 GB · fits

// budgets shown at ctx 8K, concurrency 1, 15% safety headroom. Tune in the calculator →

Alternatives at this size.

$ grep --params similar catalog.json
$ similarStarCoder2 15B15BBigCode$ similarQwen 3.5 9B9BAlibaba$ similarGemma 2 9B9BGoogle$ similargpt-oss 20B20.9BOpenAI

Discussion.

$ gh discussion list

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