Yes — with 34.5 GB to spare
Qwen3.8 27B at Q4_K_M fits your CPU only · DDR5 dual-channel entirely on the GPU at 8K context, at an estimated 2.4 tokens per second. There is room for its full 256K window.
Fully on GPU
8K context
Q4_K_M · 15.6 GB
Apache 2.0
Released 14 Aug 2026
New this week
Vision
The current default local Qwen: dense 27B, text + image + video, 262K context. Only 16 of its 64 blocks keep a KV cache, so long context is cheap.
The VRAM budget
weights 15.6 GB
Weights 15.6 GB
KV cache @ 8K 0.50 GB
Runtime overhead 0.6 GB
Free 34.5 GB of 51.2 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| Q8_0 | 27.5 GB | 28.6 GB | 256K | 1.4 | −0.1% ppl | Long context |
| Q6_K | 21.2 GB | 22.3 GB | 256K | 1.8 | −0.4% ppl | Long context |
| Q5_K_M | 18.4 GB | 19.5 GB | 256K | 2.1 | −0.8% ppl | Long context |
| Q4_K_M | 15.6 GB | 16.7 GB | 256K | 2.4 | −1.9% ppl | Recommended |
| Q3_K_M | 12.7 GB | 13.8 GB | 256K | 3.0 | −5.4% ppl | Long context |
| Q2_K | 10.8 GB | 11.9 GB | 256K | 3.5 | −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. Only 16 of its 64 blocks keep a per-token KV cache; the rest are linear-attention, Mamba or convolution blocks with a fixed-size state.
How to run it
$ llama-server \
-hf Qwen/Qwen3.8-27B:Q4_K_M \
-c 8192 -ngl 99
The engine underneath most of the others. Every knob is exposed. More on llama.cpp.
01Download is 15.6 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02With no GPU, thread count matters more than clock. Start at one thread per physical core.
03There is room to go to the model's full 256K context on this card.