LLM
How to deploy phi-4 on a GPU cloud
A 14.7B language model for chat and instruction-following. Full specs, license and use cases.
phi-4 size and hardware requirements
14.7B
Total parameters
Dense (no MoE)
Architecture
BF16
Published precision
32.8 GB
Min VRAM (native)
| Precision | Weight size on disk | Required VRAM | Cheapest live fit | GPUs needed | Est. $/hr (full fit) |
|---|---|---|---|---|---|
| BF16 | 27.3 GB | 32.8 GB | RTX A6000 | 1 | $0.330/hr |
| FP8 (quantized) | 13.7 GB | 16.4 GB | RTX 4000 SFF Ada | 1 | $0.180/hr |
| INT4 (quantized) | 6.8 GB | 8.2 GB | RTX 4070 Super | 1 | $0.110/hr |
How to run phi-4
Run phi-4 with vLLM
Generic example, not from the model's own docs: adjust flags (quantization, context length, parallelism) for your setup.
vllm serve microsoft/phi-4 --tensor-parallel-size 1Run phi-4 with Ollama
Verified against Ollama's own library listing.
ollama run phi4Source: https://ollama.com/library/phi4
Run phi-4 with GGUF quantizations
Prebuilt GGUF weights published at bartowski/phi-4-GGUF. Run with llama.cpp's llama-server or load the repo directly in LM Studio.
llama-server -hf bartowski/phi-4-GGUFDeploy phi-4 on Aquanode
Aquanode has no one-click deploy template for phi-4; you install the inference engine yourself with the commands below. Aquanode sells GPU pods billed per second, not a hosted inference API.
- Launch a bare GPU pod sized to the requirement above (1× RTX A6000 or larger).
- Open a terminal on the pod, or save one of the commands above as a startup script so it runs automatically the first time the pod boots.
- Run the command and connect to the resulting endpoint.
Launch a GPU pod