How to Deploy Qwen3.8 27B Locally
Qwen3.8 27B is a dense model with 27.8B parameters. At Q4 it needs at least 19GB of VRAM, and 24GB is the comfortable amount for an 8K context. The cheapest card that runs it today is the GeForce RTX 2080 Ti 22GB at about $387, at roughly 25.1 tokens/s.
Qwen3.8 27B details
How much VRAM does Qwen3.8 27B need for local deployment?
| Quantization | WeightsGB | Total at 8KGB | Total at 32KGB | Total at 128KGB |
|---|---|---|---|---|
| 1-bit | 6.7GGUF | 9.5 | 15.5 | 39.5 |
| 2-bit | 9.8GGUF | 13.3 | 19.3 | 43.3 |
| 3-bit | 13.1GGUF | 16.3 | 22.3 | 46.3 |
| 4-bit | 16.5GGUF | 18.5 | 24.5 | 48.5 |
| 5-bit | 19.8GGUF | 21.7 | 27.7 | 51.7 |
| 8-bit | 29GGUF | 31.8 | 37.8 | 61.8 |
| FP16 | 54.7GGUF | 57.3 | 63.3 | 87.3 |
Total = weights + KV cache (FP16) + 1GB runtime overhead. Weights marked GGUF are measured file sizes from the repository; the rest are estimated from bytes per parameter. Contexts beyond this model's 256K limit show a dash. Each tier takes whichever file of that bit width the repository has, preferring unsloth dynamic quants, so the exact file name varies by model.
Which GPUs can run Qwen3.8 27B?
Xianyu
| Device | VRAMGB | Price | Est. t/s | Suggested context |
|---|---|---|---|---|
 GeForce RTX 2080 Ti 22GB | 22 | $3873 listings | 25.1 | 8K |
 Instinct MI50 32GB | 32 | $4054 listings | 29.7 | 32K |
| Tesla V100 32GB | 32 | $47611 listings | 36.3 | 32K |
| Radeon RX 7900 XT | 20 | $63916 listings | 23.3 | 8K |
 Arc Pro B60 | 24 | $78410 listings | 12.1 | 8K |
| Radeon RX 7900 XTX | 24 | $86217 listings | 27.9 | 8K |
| GeForce RTX 3090 | 24 | $1,20415 listings | 37.8 | 8K |
| Arc Pro B70 | 32 | $1,4656 listings | 16 | 32K |
| Radeon AI PRO R9700 | 32 | $1,6438 listings | 18.7 | 32K |
| CMP 170HX 40GB | 40 | $1,6804 listings | 61.7 | 32K |
| Radeon PRO W7800 | 32 | $1,8443 listings | 16.9 | 32K |
| CMP 170HX 64GB | 64 | $2,0898 listings | 59.1 | 128K |
| RTX PRO 4000 Blackwell | 24 | $2,32718 listings | 27.3 | 8K |
| Instinct MI210 | 64 | $2,75112 listings | 46.8 | 128K |
| Radeon PRO W7900 | 48 | $3,2715 listings | 25.1 | 32K |
| GeForce RTX 4090 | 24 | $3,49416 listings | 40.6 | 8K |
| GeForce RTX 5090 D V2 | 24 | $3,55414 listings | 53.5 | 8K |
| RTX PRO 4500 Blackwell | 32 | $3,84416 listings | 36.2 | 32K |
| GeForce RTX 4090 48GB | 48 | $3,9553 listings | 40.6 | 32K |
| A100 40GB PCIe | 40 | $4,6024 listings | 61.5 | 32K |
| Mac Studio M4 Max 128GB | 128 | $4,78813 listings | 17.6 | 128K |
| DGX Spark 128GB | 128 | $4,83231 listings | 11.2 | 128K |
| GeForce RTX 5090 D | 32 | $4,89213 listings | 70.3 | 32K |
| MacBook Pro M4 Max 128GB | 128 | $5,2644 listings | 17.6 | 128K |
| Mac Studio M5 Max 128GB | 128 | $5,8816 listings | 19.7 | 128K |
| GeForce RTX 5090 | 32 | $6,51313 listings | 70.3 | 32K |
| RTX PRO 5000 Blackwell 48GB | 48 | $6,95915 listings | 53.5 | 32K |
| MacBook Pro M5 Max 128GB | 128 | $8,1487 listings | 19.7 | 128K |
| RTX PRO 5000 Blackwell 72GB | 72 | $9,66517 listings | 53.5 | 128K |
| RTX PRO 6000D | 84 | $10,11117 listings | 62 | 128K |
| RTX PRO 6000 Blackwell | 96 | $18,4226 listings | 70.3 | 128K |
| Mac Studio M3 Ultra 512GB | 512 | $19,3156 listings | 26.2 | 128K |
| Arc Pro B65 | 32 | —No listings | 16 | 32K |
| GeForce RTX 4080 SUPER 32GB | 32 | —No listings | 29.9 | 32K |
| Instinct MI100 | 32 | —No listings | 35.5 | 32K |
| Ryzen AI Max+ 395 128GB | 128 | —No listings | 7.6 | 128K |
Fits when weights + KV cache + runtime overhead is at or below usable memory, single card. "Offload" on MoE models means it runs with expert weights in system RAM (only attention layers and KV cache stay in VRAM); that speed assumes 70 GB/s RAM bandwidth. The context column is the largest tier that fits.
Deploy Qwen3.8 27B locally with Ollama, llama.cpp or vLLM
The Ollama library has no official tag for it yet.
llama.cpp pulls the GGUF straight from Hugging Face; -c sets the context length.
llama-server -hf unsloth/Qwen3.8-27B-GGUF:Q4_K_M -c 8192
vLLM serves the original-precision weights, which needs far more memory than GGUF and usually more than one GPU.
vllm serve Qwen/Qwen3.8-27B
File names follow that week's model-library snapshot; the definitions are in the methodology. Methodology
FAQ
How much VRAM does Qwen3.8 27B need?
At Q4_K_M with an 8K context Qwen3.8 27B needs about 19GB of VRAM; 24GB leaves comfortable headroom.
What is the cheapest GPU that runs Qwen3.8 27B?
The GeForce RTX 2080 Ti 22GB: 22GB of VRAM, currently about $387, at roughly 25.1 tokens/s.
Can you run Qwen3.8 27B with Ollama?
Not yet: the Ollama library has no official tag for it. You can load the GGUF with llama.cpp instead.