text model · Phi-3.5 · Windows
Can I run Phi-3.5-mini 3.8B on Nvidia GeForce RTX 4080 (16GB)?
Yes. Phi-3.5-mini 3.8B runs on Nvidia GeForce RTX 4080 (16GB) at Q4_K_M (~3.7 GB of ~15 GB usable).
Runs at Q4_K_M using ~3.7 GB of ~15 GB usable. You have room for FP16 for higher quality.
That figure is at a 4k context and moves about ±15% as context length changes. Nvidia GeForce RTX 4080 (16GB) leaves ~11.3 GB of headroom, room to step up to FP16 for higher quality.
- Q4_K_M needed
- ~3.7 GB
- Usable on device
- ~15 GB
- Device memory
- 16 GB
- Best quant
- Q4_K_M
Which quant fits
Running cost · estimate
- Power draw
- ~320 W
- Electricity / 1M tokens
- ~$0.07
- Pays for itself after
- ~2,788M tok
At ~$0.15/kWh and the estimated ~195 tok/s, a million generated tokens costs about $0.07 in electricity. Versus a hosted API at ~$0.5 per million tokens, the ~$1,199 Nvidia GeForce RTX 4080 (16GB) pays for itself after roughly 2,788 million tokens, so local hardware is mostly a fixed cost, not a per-token one. TDP is the peak draw, so this is an upper bound. Assumptions.
Run it
Pick your tool. All 3 load the same Q4_K_M weights.
ollama run phi3.5:3.8b llama-cli -hf bartowski/Phi-3.5-mini-instruct-GGUF:Q4_K_M lms get bartowski/Phi-3.5-mini-instruct-GGUF - Parameters
- 3.82B
- Q4_K_M size
- 2.39 GB
- Q8_0 size
- 4.06 GB
- Context
- 128k
- Ollama tag
- phi3.5:3.8b
- Memory
- 16 GB vram
- Usable for weights
- ~15 GB
- Power draw
- ~320 W
- Best runtime
- vLLM (Linux) / Ollama (CUDA)
You could also run
Run Phi-3.5-mini 3.8B on other hardware
FAQ
Can Nvidia GeForce RTX 4080 (16GB) run Phi-3.5-mini 3.8B?
Yes. Phi-3.5-mini 3.8B runs on Nvidia GeForce RTX 4080 (16GB) at Q4_K_M (~3.7 GB of ~15 GB usable).
How much memory does Phi-3.5-mini 3.8B need?
Nvidia GeForce RTX 4080 (16GB) has room to spare. At Q4_K_M the weights are ~2.39 GB; with KV cache and runtime overhead, budget ~3.7 GB at a 4k context.
What is the best tool to run Phi-3.5-mini 3.8B 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/phi-3.5-mini/nvidia-rtx-4080-16gb) 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.