StarCoder2 on NVIDIA GeForce RTX 4090 · 24GB VRAM

✅ Runs comfortably — 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 21.6 GB usable VRAM — plenty of headroom. Expect ~25–40 tok/s (fast).
context @ Q8_0: comfortable to 8K · runs to its full 16K
quantneedsspeed
✓ Q8_017.2 GB~25–40 tok/s (fast)
✓ Q5_K_M11.7 GB~35–60 tok/s (fast)
✓ Q4_K_M10.1 GB~40–70 tok/s (very fast)
✓ Q3_K_M8.4 GB~50–85 tok/s (very fast)
$runlocal install starcoder2:15b
⚠ NVIDIA support is best-effort in v0.1 — verify before relying on it.
runlocal verdict card
download card share on x share on reddit drop it in your model-drop post
StarCoder2: 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 NVIDIA GeForce RTX 4090: DeepSeek-R1 (distill) · Qwen3 · Qwen2.5 · Gemma 3 · Gemma 4

Best models for a NVIDIA GeForce RTX 4090 · 24GB VRAM →

Architecturally similar

These share StarCoder2's KV-cache geometry — they size memory the same way, so here's how they fit on a NVIDIA GeForce RTX 4090 · 24GB VRAM 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 →