Latest Content
Phi 4 Mini Reasoning icon

Phi 4 Mini Reasoning

NVIDIA
Phi 4 Mini Reasoning is a lightweight dense transformer model optimized for multi-step mathematical and logical reasoning with agentic tool-calling capabilities in constrained environments. It features 3.8B parameters with 32 layers, 3,072 hidden size, and 24 attention heads using grouped-query attention with 8 key-value heads. The model supports a 128K token context window with long RoPE scaling and a sliding window of 262K tokens, along with a 200K vocabulary. Trained on ~150B tokens, it delivers efficient, low-latency structured problem solving.
TypeDense LLM
CapabilitiesText Generation, Instruction Following, Reasoning, Mathematical Reasoning+5 more
Release Date30 April, 2025
Links
Paper|Blog|HF Model Card
LicenseMIT

Inference Instructions

Deploy and run this model on NVIDIA B200 GPUs using the command below. Copy the command to get started with inference.

CONSOLE
docker run -it --rm 
 --runtime=nvidia 
 --gpus all 
 --ipc=host 
 --shm-size=128g 
 -p 8000:8000 
 -v ~/.cache/huggingface:/root/.cache/huggingface 
 -e HF_TOKEN='YOUR_HF_TOKEN' 
 vllm/vllm-openai:v0.18.1 
 microsoft/Phi-4-mini-reasoning 
  --tensor-parallel-size 8 
 --max-model-len auto 
 --enable-auto-tool-choice 
 --tool-call-parser phi4_mini_json 
 --gpu-memory-utilization 0.95 
 --max-num-batched-tokens 65536 
 --max-num-seqs 1024 
 --trust-remote-code

Model Benchmarks

Each model was tested with a fixed input size and total token volume while increasing concurrency to measure serving performance under load.

ITL vs Concurrency

Time to First Token

Throughput Scaling

Total Tokens/sec vs Avg TTFT

Vultr Cloud GPU

NVIDIA HGX B200

Deploy NVIDIA B200 on Vultr Cloud GPU

Deploy NVIDIA B200 instancesContact Sales