Yes — with 180.0 GB to spare
Qwen2.5-Coder 14B at Q4_K_M fits your Instinct MI300X entirely on the GPU at 8K context, at an estimated 386 tokens per second. There is room for its full 128K window.
Fully on GPU
8K context
Q4_K_M · 8.3 GB
Apache 2.0
Released Nov 2024
Noticeably better at whole-file edits than the 7B, still comfortable on 12 GB.
The VRAM budget
weights 8.3 GB
Weights 8.3 GB
KV cache @ 8K 1.50 GB
Runtime overhead 0.6 GB
Free 180.0 GB of 190.4 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| F16 | 27.6 GB | 29.7 GB | 128K | 116 | Reference | Long context |
| Q8_0 | 14.6 GB | 16.7 GB | 128K | 219 | −0.1% ppl | Long context |
| Q6_K | 11.3 GB | 13.4 GB | 128K | 284 | −0.4% ppl | Long context |
| Q5_K_M | 9.8 GB | 11.9 GB | 128K | 328 | −0.8% ppl | Long context |
| Q4_K_M | 8.3 GB | 10.4 GB | 128K | 386 | −1.9% ppl | Recommended |
| Q3_K_M | 6.7 GB | 8.8 GB | 128K | 476 | −5.4% ppl | Long context |
| Q2_K | 5.8 GB | 7.9 GB | 128K | 556 | −15% ppl | Long context |
Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it.
How to run it
$ ollama pull qwen2.5-coder:14b $ OLLAMA_CONTEXT_LENGTH=8192 \ ollama run qwen2.5-coder:14b
The default. One binary, a model registry, an OpenAI-compatible port. More on Ollama.
01Download is 8.3 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02Close anything else holding VRAM. A browser with hardware acceleration can sit on 1–2 GB.
03There is room to go to the model's full 128K context on this card.