@meta
  v: 1
  route: /gpu/m2-max-64
  generated: 2026-06-10T09:08:08.686Z

@intent
  purpose:    Show what AI models fit on the M2 Max 64.
  audience:   ai-engineer, self-hoster, homelab-buyer
  capability: inspect_specs, find_fitting_models, compare_to_other_gpus

@state
  slug: m2-max-64
  name: M2 Max 64
  vram_gb: 64
  bandwidth_gbs: 400
  fp16_tflops: 27
  budget_8k_gb: 50
  category: apple
  summary: "Unified 64GB. Big models, slow speed. 70B at Q5, but tokens-per-second is the trade."
  top_fitting_models[5]: command-r-plus, qwen-2-5-72b, llama-3-3-70b, mixtral-8x7b, qwen-3-6-35b-a3b

@actions
  - id: compare
    method: GET
    href: /gpu/m3-max-64
  - id: compare
    method: GET
    href: /gpu/m4-pro-64
  - id: compare
    method: GET
    href: /gpu/m5-pro-64
  - id: compare
    method: GET
    href: /gpu/h100
  - id: view_calculator
    method: GET
    href: /#calculator
  - id: view_math
    method: GET
    href: /the-math

@context
  > Unified 64GB. Big models, slow speed. 70B at Q5, but tokens-per-second is the trade.

@nav
  self:      /gpu/m2-max-64
  parents:   [/]
  peers:     [/gpu/m3-max-64, /gpu/m4-pro-64, /gpu/m5-pro-64, /gpu/h100]
  drilldown: /the-math