Datagram

Run Llama 3.1 8B on a Dedicated Mac

Llama 3.1 8B needs about 10.0 GB of memory at 4-bit. It runs on our Mac Mini M6 32GB at about 13–19 tokens per second, from $169/mo.

Macs that run Llama 3.1 8B

At 4-bit with a 32K-token context. Speeds are estimates; see real measurements on the calculator page.

  • Mac Mini M6 32GB Best value
    13–19 tokens/sec
    from $169/mo
    Order
  • Mac Mini M5 Pro 64GB
    23–33 tokens/sec
    from $299/mo
    Order
  • Mac Studio M4 Max 128GB
    39–54 tokens/sec
    from $499/mo
    Order
  • Mac Studio M5 Max 128GB
    44–60 tokens/sec
    from $569/mo Available Nov 10, 2026
    Order
  • Mac Studio M5 Ultra 256GB
    77–101 tokens/sec
    from $1,199/mo Available Nov 16, 2026
    Order
  • 2× Mac Studio M5 Ultra 256GB Cluster Cluster
    67–84 tokens/sec
    from $2,399/mo Available Nov 16, 2026
    Order
  • Mac Studio M5 Ultra 512GB
    77–101 tokens/sec
    from $2,499/mo Available Dec 30, 2026
    Order
  • 2× Mac Studio M5 Ultra 512GB Cluster Cluster
    67–84 tokens/sec
    from $4,999/mo Available Dec 30, 2026
    Order
  • 4× Mac Studio M5 Ultra 512GB Cluster Cluster
    53–63 tokens/sec
    from $9,999/mo Available Dec 30, 2026
    Order

Why it runs well on a Mac

Language models have to sit entirely in memory, and generating each word means reading through them again. Apple Silicon puts up to 512 GB of fast unified memory on one chip that the GPU can use directly, so a model like this fits on one quiet machine instead of a stack of graphics cards.

Llama 3.1 8B is a dense model: all 8 billion parameters are used for every word, so memory bandwidth decides its speed. The M5 Ultra's 1,229 GB/s makes the biggest difference here.

Similar models