vLLM/Recipes
Microsoft

microsoft/Phi-4-reasoning

Microsoft Phi-4 14B dense reasoning model with 32K context.

Validated on Intel Xeon 6 with BF16 serving

View on HuggingFace
dense14B32,768 ctxvLLM 0.4.2+text
Guide

Overview

Phi-4-reasoning is Microsoft's 14B dense reasoning model. Intel Xeon 6 CPU serving was validated with BF16.

Prerequisites

  • Hardware: Intel Xeon 6/Xeon 5 CPUs
  • vLLM >= 0.4.2

Docker (Intel Xeon 6 CPUs)

docker pull vllm/vllm-openai-cpu:latest-x86_64

Intel Xeon 6

vllm serve microsoft/Phi-4-reasoning \
  --tensor-parallel-size 1

Docker (the image entrypoint is vllm serve):

docker run \
  --privileged --ipc=host -p 8000:8000 \
  -v ~/.cache/huggingface:/root/.cache/huggingface \
  vllm/vllm-openai-cpu:latest-x86_64 microsoft/Phi-4-reasoning \
  --tensor-parallel-size 1

The example uses TP=1. Adjust TP/DP for the deployment topology; host-specific CPU placement is intentionally not hard-coded.

Runtime and Platform Tuning

The following settings are intentionally not prescribed as portable CPU model defaults because they depend on the workload, hardware, or runtime environment:

  • --max-num-batched-tokens: scheduler/throughput tuning.
  • --max-num-seqs: concurrency and scheduler-capacity tuning.
  • --gpu-memory-utilization: platform memory-budget tuning.
  • --no-enable-prefix-caching: workload/benchmark cache-policy tuning.
  • VLLM_ENGINE_ITERATION_TIMEOUT_S: operational runtime timeout.

These settings may still appear in validated hardware-specific overrides. Tune them at deployment time based on platform resources, workload shape, and latency/throughput goals.

References