Qwen3 on Apple M4 Max · 128GB

✅ Runs comfortably — Qwen3 32B @ Q8_0
Qwen3 32B at Q8_0 needs ~36.0 GB (weights 32.4 GB + KV 2.0 GB + overhead 1.6 GB @ 8K ctx) of your 93.0 GB usable (of 128 GB unified memory) — plenty of headroom. Expect ~6–11 tok/s (usable).
context @ Q8_0: comfortable to 128K
quantneedsspeed
✓ Q8_036.0 GB~6–11 tok/s (usable)
✓ Q5_K_M24.8 GB~9–16 tok/s (usable)
✓ Q4_K_M21.4 GB~11–18 tok/s (usable)
✓ Q3_K_M17.7 GB~13–20 tok/s (usable)
$runlocal install qwen3:32b
runlocal verdict card
download card share on x share on reddit drop it in your model-drop post
Qwen3: Runs comfortably
for your own README — links back here

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

Also runs on a Apple M4 Max: DeepSeek-R1 (distill) · Qwen2.5 · Llama 3.1 · Llama 3.3 · Mistral Small 3

Best models for a Apple M4 Max · 128GB →

Architecturally similar

These share Qwen3's KV-cache geometry — they size memory the same way, so here's how they fit on a Apple M4 Max · 128GB too:

modelsharesverdict here
Qwen2.5-Coder 4kv/128hd8kv/128hd ✅ Runs comfortably
Llama 3.2 8kv/128hd ✅ Runs comfortably
Mistral 7B 8kv/128hd ✅ Runs comfortably
Mistral Nemo 8kv/128hd ✅ Runs comfortably
Codestral 8kv/128hd ✅ Runs comfortably
Llama 3.2 Vision 8kv/128hd ✅ Runs comfortably
Mixtral 8x7B 8kv/128hd ✅ Runs comfortably
Phi-4-mini 8kv/128hd ✅ Runs comfortably

…and 10 more — see the Qwen3 page.

Check any combo yourself: open the checker →