StarCoder2 on NVIDIA GeForce RTX 5080 · 16GB VRAM
✅ Runs (tight) — StarCoder2 15B @ Q5_K_M
StarCoder2 15B at Q5_K_M needs ~11.7 GB (weights 10.6 GB + KV 640 MB + overhead 541 MB @ 8K ctx) of your 14.4 GB usable VRAM — fits, but little headroom — close other apps or trim context. Expect ~35–55 tok/s (fast).
context @ Q5_K_M: comfortable to 4K · runs to its full 16K
| quant | needs | speed |
|---|---|---|
| ~ Q8_0 | 17.2 GB | ~7–12 tok/s (usable) |
| ✓ Q5_K_M | 11.7 GB | ~35–55 tok/s (fast) |
| ✓ Q4_K_M | 10.1 GB | ~40–65 tok/s (very fast) |
| ✓ Q3_K_M | 8.4 GB | ~50–80 tok/s (very fast) |
$runlocal install starcoder2:15b
⚠ NVIDIA support is best-effort in v0.1 — verify before relying on it.
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 5080: DeepSeek-R1 (distill) · Qwen2.5 · Qwen3 · Gemma 4 · Gemma 3
Best models for a NVIDIA GeForce RTX 5080 · 16GB 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 5080 · 16GB VRAM too:
| model | shares | verdict here |
|---|---|---|
| Qwen2.5-Coder | 2kv/128hd4kv/128hd | ✅ Runs (tight) |
| Yi 1.5 | 4kv/128hd | ✅ Runs comfortably |
| Qwen3 Coder 30B-A3B | 4kv/128hd | ⚠️ Partial GPU offload |
| SmolLM3 3B | 4kv/128hd | ✅ Runs comfortably |
| Yi-Coder 9B | 4kv/128hd | ✅ Runs comfortably |
Check any combo yourself: open the checker →