Llama 3.2 on Apple M1 · 8GB

✅ Runs comfortably — Llama 3.2 1B @ Q8_0
Llama 3.2 1B at Q8_0 needs ~2.0 GB (weights 1.2 GB + KV 256 MB + overhead 512 MB @ 8K ctx) of your 2.9 GB usable (of 8.0 GB unified memory) — plenty of headroom. Expect ~25–45 tok/s (fast).
context @ Q8_0: comfortable to 16K · runs to its full 128K
quantneedsspeed
~ FP163.1 GB~2–3 tok/s (slow)
Q8_02.0 GB~25–45 tok/s (fast)
Q4_K_M1.5 GB~40–65 tok/s (very fast)
$runlocal install llama3.2:1b
runlocal verdict card
download card share on x share on reddit drop it in your model-drop post
Llama 3.2: Runs comfortably
for your own README — links back here

Same model, other machines: Apple M1 Pro · Apple M2 · Apple M3 Pro · Apple M2 Pro · Apple M4 Pro · Apple M3 Max

Also runs on a Apple M1: DeepSeek-R1 (distill) · Qwen2.5 · Qwen3 · Gemma 2 · Gemma 3

Best models for a Apple M1 · 8GB →

Architecturally similar

These share Llama 3.2's KV-cache geometry — they size memory the same way, so here's how they fit on a Apple M1 · 8GB too:

modelsharesverdict here
Llama 3.1 8kv/128hd 🐢 CPU-only (slow)
Qwen2.5-Coder 8kv/128hd ✅ Runs comfortably
Llama 3.3 8kv/128hd ❌ Won't fit
Mistral 7B 8kv/128hd 🐢 CPU-only (slow)
Mistral Small 3 8kv/128hd ❌ Won't fit
Mistral Nemo 8kv/128hd ❌ Won't fit
Codestral 8kv/128hd ❌ Won't fit
Llama 3.2 Vision 8kv/128hd ❌ Won't fit

…and 8 more — see the Llama 3.2 page.

Check any combo yourself: open the checker →