StarCoder2 on Apple M3 Max · 36GB

✅ Runs (tight) — StarCoder2 15B @ Q8_0
StarCoder2 15B at Q8_0 needs ~17.2 GB (weights 15.8 GB + KV 640 MB + overhead 809 MB @ 8K ctx) of your 21.1 GB usable (of 36.0 GB unified memory) — fits, but little headroom — close other apps or trim context. Expect ~10–17 tok/s (usable).
context @ Q8_0: comfortable to 4K · runs to its full 16K
quantneedsspeed
Q8_017.2 GB~10–17 tok/s (usable)
Q5_K_M11.7 GB~16–25 tok/s (fast)
Q4_K_M10.1 GB~18–30 tok/s (fast)
Q3_K_M8.4 GB~20–35 tok/s (fast)
$runlocal install starcoder2:15b
runlocal verdict card
download card share on x share on reddit drop it in your model-drop post
StarCoder2: Runs (tight)
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 M3 Max: Qwen3 · DeepSeek-R1 (distill) · Qwen2.5 · Gemma 4 · Gemma 3

Best models for a Apple M3 Max · 36GB →

Architecturally similar

These share StarCoder2's KV-cache geometry — they size memory the same way, so here's how they fit on a Apple M3 Max · 36GB too:

modelsharesverdict here
Qwen2.5-Coder 2kv/128hd4kv/128hd ✅ Runs (tight)
Yi 1.5 4kv/128hd ✅ Runs (tight)
Qwen3 Coder 30B-A3B 4kv/128hd ✅ Runs (tight)
SmolLM3 3B 4kv/128hd ✅ Runs comfortably
Yi-Coder 9B 4kv/128hd ✅ Runs comfortably

Check any combo yourself: open the checker →