PH

Phi4 MoE 2x14B Instruct

PhiGGUF by mradermacherUpdated 2025-02-24
Download on Hugging Face
Parameters14.2B
Context16K
Layers40
Architecturellama
Smallest quant24.2 GB
Quick answer

Phi4 MoE 2x14B Instruct needs about 26.9 GB of memory at Q4_K_M (a 24.8 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 10 quants, from 24.2 GB (Q2_K) to 25.2 GB (Q6_K).

Phi4 MoE 2x14B Instruct is a 14.2B-parameter Phi model in GGUF format, quantized by mradermacher. Its trained context length is 16K tokens. The repository has 10 quant files, from 24.2 GB (Q2_K) to 25.2 GB (Q6_K).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K24.2 GB 26.3 GB
Too big Get file →
Q3_K_S24.3 GB 26.5 GB
Too big Get file →
Q3_K_M24.4 GB 26.6 GB
Too big Get file →
Q3_K_L24.5 GB 26.7 GB
Too big Get file →
IQ4_XS24.6 GB 26.8 GB
Too big Get file →
Q4_K_S24.7 GB 26.9 GB
Too big Get file →
Q4_K_M24.8 GB 26.9 GB
Too big Get file →
Q5_K_S24.9 GB 27.1 GB
Too big Get file →
Q5_K_M25.0 GB 27.1 GB
Too big Get file →
Q6_K25.2 GB 27.4 GB
Too big Get file →

What GPU do you need for Phi4 MoE 2x14B Instruct?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: no quant fits at an 8K context.
  • 12 GB VRAM: no quant fits at an 8K context.
  • 16 GB VRAM: no quant fits at an 8K context.
  • 24 GB VRAM: no quant fits at an 8K context.
  • 32 GB VRAM: the best quant that fits is Q6_K (25.2 GB file, about 27.4 GB in use).

Which Phi4 MoE 2x14B Instruct quant should you choose?

Q4_K_M is the usual starting point for Phi4 MoE 2x14B Instruct: it balances file size and quality well. Here it is a 24.8 GB file.

If memory is tight, the smallest file is Q2_K at 24.2 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturellama
Parameters14.2B (approximate, mixture of experts)
Layers40
Trained context16,384 tokens
KV cache (fp16)200 KB per token

Source

Quantized bymradermacher
FamilyPhi
Licenseother
Downloads (HF)267
Last updated2025-02-24

Frequently asked questions about Phi4 MoE 2x14B Instruct

How much VRAM does Phi4 MoE 2x14B Instruct need?

About 26.9 GB at Q4_K_M with an 8K context window. That is the 24.8 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run Phi4 MoE 2x14B Instruct on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 26.3 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Phi4 MoE 2x14B Instruct GGUF?

It depends on the quant: Q2_K is 24.2 GB, Q3_K_S is 24.3 GB, Q3_K_M is 24.4 GB, Q3_K_L is 24.5 GB, IQ4_XS is 24.6 GB, Q4_K_S is 24.7 GB and 4 more.

What context length does Phi4 MoE 2x14B Instruct support?

The GGUF header lists a trained context length of 16,384 tokens (16K). The KV cache grows with the context you use: about 1.7 GB at 8K and 3.4 GB at 16K.

Where can I download Phi4 MoE 2x14B Instruct GGUF?

The files are on Hugging Face in the mradermacher/Phi4-MoE-2x14B-Instruct-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Phi4 MoE 2x14B Instruct use?

Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Phi models

All Phi →

Other medium models

All medium models →