Apple M1, 8 GB · Llama 3.2 1B Instruct · Q4_K_M
Yes. Apple M1, 8 GB runs Llama 3.2 1B Instruct at Q4_K_M with room left at 8,192 tokens, roughly 35–67 tokens per second.
| Piece | Value | Note |
|---|---|---|
| Weights | 0.75 GiB | Llama-3.2-1B-Instruct-Q4_K_M.gguf |
| KV cache | 0.25 GiB | 8,192 tokens |
| Engine overhead | 0.50 GiB | Flat 0.5 GiB allowance |
| 5% headroom | 0.08 GiB | On the sum above |
| Total | 1.58 GiB | Fits |
| Fast memory | 5.00 GiB | After OS reserve |
| Speed | 35–67 tok/s | Dense bandwidth range |
ollama run llama3.2:1b
Apple M1, 8 GB · Llama 3.2 1B Instruct ·GGUF file · Methodology
FAQ
How was this fit calculated?
Weights 0.75 GiB, KV cache 0.25 GiB at 8,192 tokens, engine overhead 0.50 GiB, then 5% headroom, for 1.58 GiB. Fast memory available is 5.00 GiB.
What if I change the quant or the context?
This page is Q4_K_M at 8,192 tokens. A smaller quant or a shorter context lowers the total. The homepage calculator changes both without a new page.
Can I trust the tokens per second?
The range 35–67 tok/s is a bandwidth roofline, 55% to 80% efficient. This dense model reads the whole weight file each token. Measure your own box before you buy around it.
What do I type?
ollama run llama3.2:1b
Is Apple M1, 8 GB assumed to have extra RAM?
No separate RAM pool. The chip's unified memory is the whole budget, minus the OS reserve described on the methodology page.