@meta
  v: 1
  route: /model/phi-4-mini
  generated: 2026-06-10T08:48:06.340Z

@intent
  purpose:    Show what hardware can run Phi-4 Mini.
  audience:   ai-engineer, self-hoster, model-evaluator
  capability: inspect_model_specs, find_compatible_gpus, compare_similar_models, open_huggingface

@state
  slug: phi-4-mini
  name: Phi-4 Mini
  hf_repo: microsoft/Phi-4-mini-instruct
  params_b: 3.8
  active_params_b: ~
  family: Microsoft
  type: dense
  context_k: 128
  fp16_gb: 7.6
  summary: 3.8B Phi-4 distill. Strong reasoning at edge sizes; 128K context.
  smallest_fitting_gpu_q4: rtx-3070
  smallest_fitting_gpu_fp16: rtx-3060

@actions
  - id: open_huggingface
    method: GET
    href: https://huggingface.co/microsoft/Phi-4-mini-instruct
  - id: view_compatible_gpus
    method: GET
    href: /model/phi-4-mini
  - id: compare_similar
    method: GET
    href: /model/gemma-4-e4b
  - id: view_calculator
    method: GET
    href: /#calculator

@context
  > 3.8B Phi-4 distill. Strong reasoning at edge sizes; 128K context.

@nav
  self:      /model/phi-4-mini
  parents:   [/]
  peers:     [/model/gemma-4-e4b, /model/llama-3-2-3b, /model/llama-3-2-1b, /model/qwen-2-5-7b]
  drilldown: /the-math