Phi 3.5 MoE Instruct — Hardware Requirements & GPU Compatibility
ChatCode
Phi 3.5 MoE Instruct is a 41.9B-parameter open language model from Microsoft in the Phi 3 family. It supports a context window of up to 131,072 tokens. At BF16 it needs about 84.31 GB of VRAM — see which GPUs and Macs can run it below.
138.1K downloads
574 likesSpecifications
- Publisher
- Microsoft
- Family
- Phi 3
- Parameters
- 41.9B
- Architecture
- PhiMoEForCausalLM
- Context Length
- 131,072 tokens
- Vocabulary Size
- 32,064
- License
- MIT
Get Started
HuggingFace
How Much VRAM Does Phi 3.5 MoE Instruct Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| BF16 | 16.00 | 84.3 GB | 101.2 GB | 83.75 GB | Brain floating point 16 — preferred for training |
Which GPUs Can Run Phi 3.5 MoE Instruct?
BF16 · 84.3 GBPhi 3.5 MoE Instruct (BF16) requires 84.3 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 110+ GB is recommended. Using the full 131K context window can add up to 16.9 GB, bringing total usage to 101.2 GB. No single GPU has enough memory — multi-GPU or cluster setups are needed.
Which Devices Can Run Phi 3.5 MoE Instruct?
BF16 · 84.3 GB5 devices with unified memory can run Phi 3.5 MoE Instruct, including NVIDIA DGX H100, NVIDIA DGX A100 640GB.
Benchmarks
View all 1 →Related Models
Frequently Asked Questions
- How much VRAM does Phi 3.5 MoE Instruct need?
Phi 3.5 MoE Instruct requires 84.3 GB of VRAM at BF16. Full 131K context adds up to 16.9 GB (101.2 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 41.9B × 16 bits ÷ 8 = 83.7 GB
KV Cache + Overhead ≈ 0.6 GB (at 2K context + ~0.3 GB framework)
KV Cache + Overhead ≈ 17.5 GB (at full 131K context)
VRAM usage by quantization
BF1684.3 GBBF16 + full context101.2 GB- Can NVIDIA GeForce RTX 5090 run Phi 3.5 MoE Instruct?
No — Phi 3.5 MoE Instruct requires at least 84.3 GB at BF16, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
- Can I run Phi 3.5 MoE Instruct on a Mac?
Phi 3.5 MoE Instruct requires at least 84.3 GB at BF16, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run Phi 3.5 MoE Instruct locally?
Yes — Phi 3.5 MoE Instruct can run locally on consumer hardware. At BF16 quantization it needs 84.3 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Phi 3.5 MoE Instruct?
At BF16, Phi 3.5 MoE Instruct can reach ~35 tok/s on AMD Instinct MI300X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: AMD Instinct MI300X → 5300 ÷ 84.3 × 0.55 = ~35 tok/s
Estimated speed at BF16 (84.3 GB)
~35 tok/s~21 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Phi 3.5 MoE Instruct?
At BF16, the download is about 83.75 GB.
- Which GPUs can run Phi 3.5 MoE Instruct?
No single consumer GPU has enough VRAM to run Phi 3.5 MoE Instruct at BF16 (84.3 GB). Multi-GPU or professional hardware is required.
- Which devices can run Phi 3.5 MoE Instruct?
5 devices with unified memory can run Phi 3.5 MoE Instruct at BF16 (84.3 GB), including Mac Pro M2 Ultra (192 GB), Mac Studio M2 Ultra (192 GB), Mac Studio M4 Max (128 GB), NVIDIA DGX A100 640GB. Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.