Rent a GPU

Phi-4 Reasoning 14B hardware requirements

14.7B parameters. Needs about 11.2 GB at Q4_K_M with 8K context.

Parameters15BDense
Q4_K_M · 8K11.2 GBthe usual download
FP16 · 8K31.7 GBfull precision
Runs well on40/51GPUs and Macs we track

Memory needed at each quantization

Weights, KV cache and runtime overhead, at 8K and 32K tokens of context.

QuantQuality8K context32K context
FP16Full quality31.7 GB36.7 GB
Q8_0Near-lossless17.9 GB22.9 GB
Q6_KExcellent14.3 GB19.4 GB
Q5_K_MVery good12.7 GB17.8 GB
Q4_K_MGood, most popular11.2 GB16.2 GB
Q3_K_MNoticeably degraded9.5 GB14.5 GB
Q2_KHeavily degraded7.8 GB12.8 GB

Run it with Ollama

Downloads Q4_K_M by default.

ollama run phi4-reasoning:14b

Model card

microsoft/Phi-4-reasoning on Hugging Face: licence, prompt format and GGUF builds.

Run it in the cloud

Run Phi-4 Reasoning 14B on a rented GPU today.

It needs about 11.2 GB at Q4_K_M: a rented RTX 3090 / RTX 4090 runs it for about $0.18 to $0.89 an hour.

RunPodRent a GPU on RunPodVast.aiRent a GPU on Vast.ai

Every GPU and Mac checked

Sorted by memory. Bars show how much of each card Phi-4 Reasoning 14B fills at the best quant that fits.

40 Runs well

NameMemory usedBest quantSpeed

11 Fits with heavy quantization

NameMemory usedBest quantSpeed

Other Phi models