text model · Qwen3-Coder · Windows
Can I run Qwen3-Coder-Next 80B-A3B on Nvidia GeForce RTX 3090 (24GB)?
No. Qwen3-Coder-Next 80B-A3B needs ~48 GB even at Q4_K_M, but Nvidia GeForce RTX 3090 (24GB) only has ~23 GB usable.
Needs ~48 GB even at Q4_K_M, but only ~23 GB is usable.
That figure is at a 4k context and moves about ±15% as context length changes.
The gap is about 25 GB: Qwen3-Coder-Next 80B-A3B needs roughly 48 GB at Q4_K_M and Nvidia GeForce RTX 3090 (24GB) leaves only about 23 GB usable for a model. The lightest tracked hardware that runs Qwen3-Coder-Next 80B-A3B is the Apple M4 Max (64GB) at 64 GB. See Qwen3-Coder-Next 80B-A3B on Apple M4 Max (64GB).
- Q4_K_M needed
- ~48 GB
- Usable on device
- ~23 GB
- Device memory
- 24 GB
Which quant fits
- Parameters
- 80B (MoE, 3B active)
- Q4_K_M size
- 45.09 GB
- Q8_0 size
- 78.99 GB
- Context
- 256k
- Ollama tag
- qwen3-coder-next:latest
- Memory
- 24 GB vram
- Usable for weights
- ~23 GB
- Power draw
- ~350 W
- Best runtime
- vLLM (Linux) / Ollama (CUDA)
What you can run instead
Run Qwen3-Coder-Next 80B-A3B on other hardware
FAQ
Can Nvidia GeForce RTX 3090 (24GB) run Qwen3-Coder-Next 80B-A3B?
No. Qwen3-Coder-Next 80B-A3B needs ~48 GB even at Q4_K_M, but Nvidia GeForce RTX 3090 (24GB) only has ~23 GB usable.
How much memory does Qwen3-Coder-Next 80B-A3B need?
Nvidia GeForce RTX 3090 (24GB) does not have enough memory. At Q4_K_M the weights are ~45.09 GB; with KV cache and runtime overhead, budget ~48 GB at a 4k context. It is a Mixture-of-Experts model (80B total / 3B active), so all experts must stay in memory; memory tracks total params, not active params.
What is the best tool to run Qwen3-Coder-Next 80B-A3B on Windows?
LM Studio for a simple setup; Ollama (CUDA) for the most speed. AMD GPUs run via Vulkan/ROCm at roughly half CUDA throughput. NVIDIA is the smooth path on Windows.
Embed this
A live badge for your model card or README, updated as the data is. [](https://localmodel.run/can-i-run/qwen3-coder-next/nvidia-rtx-3090-24gb) Sources
-
en.wikipedia.org · 1 source
-
hardware-corner.net · 1 source
-
huggingface.co · 2 sources
-
lmstudio.ai · 1 source
-
nvidia.com · 1 source
-
ollama.com · 2 sources
-
wccftech.com · 1 source
Weights are measured from GGUF files; KV cache and overhead are computed, so totals can vary ~15% with context and runtime. Any tok/s is a bandwidth estimate. See methodology.