text model · Command · Windows
Can I run Command A on Nvidia GeForce RTX 5090 (32GB)?
No. Command A needs ~70.4 GB even at Q4_K_M, but Nvidia GeForce RTX 5090 (32GB) only has ~31 GB usable.
Needs ~70.4 GB even at Q4_K_M, but only ~31 GB is usable.
That figure is at a 4k context and moves about ±15% as context length changes.
The gap is about 39.4 GB: Command A needs roughly 70.4 GB at Q4_K_M and Nvidia GeForce RTX 5090 (32GB) leaves only about 31 GB usable for a model. The lightest tracked hardware that runs Command A is the Apple M4 Max (128GB) at 128 GB. See Command A on Apple M4 Max (128GB).
Too big for Nvidia GeForce RTX 5090 (32GB)'s VRAM, but you could offload some layers to system RAM and still run it at roughly ~0.8 tok/s, far slower than a model that fits. How this is estimated.
- Q4_K_M needed
- ~70.4 GB
- Usable on device
- ~31 GB
- Device memory
- 32 GB
Which quant fits
- Parameters
- 111B
- Q4_K_M size
- 67.1 GB
- Q8_0 size
- 118 GB
- Context
- 256k
- Ollama tag
- command-a:111b
- Memory
- 32 GB vram
- Usable for weights
- ~31 GB
- Power draw
- ~575 W
- Best runtime
- vLLM (Linux) / Ollama (CUDA)
What you can run instead
Run Command A on other hardware
FAQ
Can Nvidia GeForce RTX 5090 (32GB) run Command A?
No. Command A needs ~70.4 GB even at Q4_K_M, but Nvidia GeForce RTX 5090 (32GB) only has ~31 GB usable.
How much memory does Command A need?
Nvidia GeForce RTX 5090 (32GB) does not have enough memory. At Q4_K_M the weights are ~67.1 GB; with KV cache and runtime overhead, budget ~70.4 GB at a 4k context.
What is the best tool to run Command A 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/command-a-111b/nvidia-rtx-5090-32gb) Sources
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.