Hardware guide · October 2026
What GPU do you need to run Qwen3.8 27B?
Qwen3.8 27B is a 27B dense model. Here is exactly how much memory it needs and which GPUs and Macs run it well.
The short answer
To run Qwen3.8 27B at Q4_K_M with 8K tokens of context you need about 18GB of VRAM (16.5GB of weights plus KV cache and overhead). The cheapest way in is a 16GB card like the RTX 4060 Ti 16GB (~$380 used). For the best speed per dollar, get the RTX 3090 (~39 tokens/sec). On a Mac or other unified-memory system, the Ryzen AI Max+ 395 runs it at ~6.4 tokens/sec.
- VRAM at Q4_K_M
- ~18GB
- GPUs that run it
- 31
- Parameters
- 27B
- CPU only (est.)
- ~2.7 tok/s
Our recommendations
The best GPUs for Qwen3.8 27B
Minimum
RTX 4060 Ti 16GB16GB · ~$380 used
- Quant
- Q3_K_M
- Speed (est.)
- ~17 tok/s
The smallest card that runs it at Q3_K_M with 8K context.
Best value
RTX 309024GB · ~$1,050 used
- Quant
- Q4_K_M
- Speed (est.)
- ~39 tok/s
The most tokens per second per dollar among cards with room to spare.
Fastest
RTX 509032GB · ~$4,300
- Quant
- Q6_K
- Speed (est.)
- ~59 tok/s
The fastest single GPU in our database for this model.
Mac / unified memory
Ryzen AI Max+ 395128GB unified · ~$3,300
- Quant
- Q8_0
- Speed (est.)
- ~6.4 tok/s
Lots of memory per dollar and silent, but slower than a GPU with the same capacity.
We may earn a commission when you buy through links on this page. It never changes what we recommend. How we make money
Memory
Qwen3.8 27B VRAM requirements
Memory needed at each quantization level. Q4_K_M is the best balance of size and quality for most people; move up to Q5 or Q6 if you have the room.
| Quantization | Quality | Weights | Total @ 8K ctx | Total @ 32K ctx |
|---|---|---|---|---|
| Q2_K | Rough | 10 GB | 12 GB | 13 GB |
| Q3_K_M | Usable | 13 GB | 15 GB | 16 GB |
| Q4_K_M | RecommendedDefault | 16 GB | 18 GB | 20 GB |
| Q5_K_M | Very good | 20 GB | 22 GB | 23 GB |
| Q6_K | Near-lossless | 23 GB | 25 GB | 26 GB |
| Q8_0 | Practically lossless | 30 GB | 32 GB | 33 GB |
| BF16 | Full precision | 56 GB | 58 GB | 60 GB |
Context
How context length changes the memory you need
Every token in the context window is stored in the KV cache. At Q4_K_M, with a 16-bit KV cache, Qwen3.8 27B needs:
| Context | KV cache | Total memory | With 8-bit KV |
|---|---|---|---|
| 4K tokens | 0.3 GB | 17.8 GB | 17.7 GB |
| 8K tokens | 0.5 GB | 18.1 GB | 17.8 GB |
| 16K tokens | 1.1 GB | 18.6 GB | 18.1 GB |
| 32K tokens | 2.1 GB | 19.7 GB | 18.6 GB |
| 64K tokens | 4.3 GB | 21.8 GB | 19.7 GB |
| 128K tokens | 8.6 GB | 26.1 GB | 21.8 GB |
| 256K tokens | 17.2 GB | 34.7 GB | 26.1 GB |
Hybrid layout: 48 of 64 layers use Gated DeltaNet linear attention with a fixed-size state, and only the 16 full-attention layers keep a growing cache: 2 x 16 x 4 KV heads x 256 dims x 2 bytes = 65,536 bytes (64 KiB) per token, about 17GB at the full 262K context.
Compatibility
Every GPU that runs Qwen3.8 27B
Each GPU at the highest-quality quantization that fits with 8K context, fastest first. Speeds are single-user estimates.
| GPU | Memory | Quant | Speed | Fit | Price |
|---|---|---|---|---|---|
| RTX 5090 | 32GB | Q6_K | ~59 tok/s | Runs well | ~$4,300 |
| RTX 4090 | 24GB | Q4_K_M | ~47 tok/s | Runs well | ~$2,500 used |
| RTX PRO 6000 | 96GB | Q8_0 | ~46 tok/s | Runs well | ~$16,000 |
| RTX 3090 | 24GB | Q4_K_M | ~39 tok/s | Runs well | ~$1,050 used |
| RX 7900 XTX | 24GB | Q4_K_M | ~36 tok/s | Runs well | ~$1,050 used |
| RTX PRO 5000 72GB | 72GB | Q8_0 | ~35 tok/s | Runs well | ~$9,100 |
| RTX 6000 Ada | 48GB | Q8_0 | ~25 tok/s | Runs well | ~$4,750 used |
| V100 32GB | 32GB | Q6_K | ~25 tok/s | Runs well | ~$700 used |
| M5 Ultra 512GB | 512GB unified | Q8_0 | ~22 tok/s | Runs well | — |
| M5 Ultra 256GB | 256GB unified | Q8_0 | ~22 tok/s | Runs well | ~$9,499 |
| MI50 32GB | 32GB | Q6_K | ~19 tok/s | Runs well | ~$600 used |
| RTX A6000 | 48GB | Q8_0 | ~18 tok/s | Runs well | ~$4,400 used |
| R9700 | 32GB | Q6_K | ~18 tok/s | Runs well | ~$1,800 |
| M3 Ultra 512GB | 512GB unified | Q8_0 | ~15 tok/s | Runs well | $9,499 |
| Arc Pro B60 | 24GB | Q4_K_M | ~15 tok/s | Runs well | ~$650 |
| Arc Pro B70 | 32GB | Q6_K | ~14 tok/s | Runs well | ~$1,300 |
| M5 Max 128GB | 128GB unified | Q8_0 | ~13 tok/s | Runs well | ~$5,100 |
| M4 Max 128GB | 128GB unified | Q8_0 | ~12 tok/s | Runs well | $3,499 |
| Tesla P40 | 24GB | Q4_K_M | ~11 tok/s | Runs well | ~$230 used |
| Ryzen AI Max+ PRO 495 | 192GB unified | Q8_0 | ~6.9 tok/s | Runs well | ~$6,800 |
| DGX Spark | 128GB unified | Q8_0 | ~6.9 tok/s | Runs well | ~$6,950 |
| Ryzen AI Max+ 395 | 128GB unified | Q8_0 | ~6.4 tok/s | Runs well | ~$3,300 |
| RTX 5080 | 16GB | Q3_K_M | ~55 tok/s | Fits, tight | ~$1,600 |
| RTX 5070 Ti | 16GB | Q3_K_M | ~51 tok/s | Fits, tight | ~$1,200 |
| RTX 4080 Super | 16GB | Q3_K_M | ~43 tok/s | Fits, tight | ~$1,150 used |
| RTX 4070 Ti Super | 16GB | Q3_K_M | ~39 tok/s | Fits, tight | ~$900 used |
| RX 9070 XT | 16GB | Q3_K_M | ~30 tok/s | Fits, tight | ~$800 |
| RTX 5060 Ti 16GB | 16GB | Q3_K_M | ~26 tok/s | Fits, tight | ~$620 |
| RTX 4000 Ada | 20GB | Q4_K_M | ~17 tok/s | Fits, tight | $1,250 |
| RTX 4060 Ti 16GB | 16GB | Q3_K_M | ~17 tok/s | Fits, tight | ~$380 used |
| RX 9060 XT 16GB | 16GB | Q3_K_M | ~15 tok/s | Fits, tight | ~$470 |
With CPU offload (partial offload, assuming 64GB of system RAM): RTX 5070 (~6.0 tok/s), RTX 4070 (~5.8 tok/s), RTX 3060 12GB (~5.4 tok/s), Arc B580 (~5.3 tok/s), RTX 5060 (~4.0 tok/s).
Tips
Getting the most out of Qwen3.8 27B
- Start at Q4_K_M. It keeps nearly all of the quality at about a third of the 16-bit size. Step up to Q5_K_M or Q6_K only if it still fits with the context you need.
- Budget memory for context. The KV cache grows with every token. Quantizing it to 8-bit roughly halves that cost with little quality loss.
- Avoid heavy CPU offload. For a dense model, every token reads every weight. Once a large share of the model sits in system RAM, speed falls off a cliff. A second GPU usually beats offloading.
- Pick the right runtime. Ollama and LM Studio are easiest. llama.cpp gives the most control. vLLM or SGLang are best when several people or agents share one machine.
Want to try different quantizations, context lengths, or GPU counts? Use our VRAM calculator, or read the Qwen3.8 27B overview for setup commands and benchmarks.
Frequently asked questions
How much VRAM do I need to run Qwen3.8 27B?
About 18GB at Q4_K_M with an 8K context window. At Q8_0 (near-lossless) it needs about 32GB, and at Q3_K_M about 15GB with some quality loss. Longer contexts add 0.07GB per 1K tokens with a 16-bit KV cache.
Can I run Qwen3.8 27B on a 16GB GPU?
Yes. A 16GB card like the RTX 5060 Ti 16GB runs it at Q3_K_M at ~55 tokens/sec.
Can I run Qwen3.8 27B on a Mac?
Yes. With 128GB of unified memory or more (for example the Ryzen AI Max+ 395), macOS can give the GPU enough memory to run it at Q8_0, at about ~6.4 tokens/sec. MLX builds are usually fastest on Apple silicon.
Can I run Qwen3.8 27B on CPU only, without a GPU?
Yes, if you have at least 32GB of system RAM. On a desktop with dual-channel DDR5, expect roughly 2.7 tokens/sec. A GPU is strongly recommended for interactive use.