# Qwen3.8 27B: RAM and VRAM requirements

> Qwen3.8 27B is a 27B Qwen3.8 model. At Q4_K_M it needs about **17.3 GB** to run and fits **13 of 43** tracked devices. Minimum to run: Nvidia GeForce RTX 4090 (24GB).

Last validated: 2026-09-07. Sources: Ollama, HuggingFace GGUF repos, vendor specs.

## Memory by quantization
| Quant | On disk | To run (4k context) |
| --- | --- | --- |
| Q4_K_M | 15.33 GB | ~17.3 GB |
| Q8_0 | 27.05 GB | ~29.1 GB |
| FP16 | 55.59 GB | ~57.6 GB |

Memory = weights + KV cache + ~0.8 GB runtime overhead, and varies ±15% with context length.

## Will it run on my device?
- **Nvidia GeForce RTX 2060 (6GB)** (6 GB): No, not enough memory
- **Generic Android Phone (8GB RAM)** (8 GB): No, not enough memory
- **Samsung Galaxy S24 Ultra** (12 GB): No, not enough memory
- **Nvidia GeForce RTX 4060 Ti (16GB)** (16 GB): No, not enough memory
- **Apple M5 (16GB)** (16 GB): No, not enough memory
- **Nvidia GeForce RTX 4090 (24GB)** (24 GB): Yes, it runs
- **Apple M4 Pro (48GB)** (48 GB): Yes, it runs (room for Q8_0)
- **Apple M3 Ultra (256GB)** (256 GB): Yes, it runs (room for FP16)

Full table of all 43 devices: https://localmodel.run/model/qwen3.8-27b

## How to run
Quickest path: `ollama run qwen3.8:27b`. On Mac, LM Studio (ships MLX) is fastest; on Linux, Ollama for chat or vLLM to serve; on Windows, LM Studio or Ollama.

## Details
- Parameters: 27B
- Default context: 256k tokens
- License: Apache-2.0 (commercial use: yes)
- Released: 2026-08
- HuggingFace: 6,416,358 downloads/mo, 14185 likes

## FAQ
### How much VRAM or RAM does Qwen3.8 27B need?
About 17.3 GB at Q4_K_M (weights 15.33 GB + KV cache + overhead) at a 4k context. Budget ~29.1 GB for Q8_0.
### Can Qwen3.8 27B run on a laptop?
Qwen3.8 27B is large; you need a high-memory Mac or a 24 GB+ GPU at Q4_K_M.
### Can I use Qwen3.8 27B commercially?
Yes, Apache-2.0 permits commercial use.

Sources: https://huggingface.co/Qwen/Qwen3.8-27B, https://huggingface.co/unsloth/Qwen3.8-27B-GGUF, https://ollama.com/library/qwen3.8/tags
More: https://localmodel.run/model/qwen3.8-27b