PH

phi4 multimodal quantisized

PhiGGUF by ShayanCyanUpdated 2026-02-16
Download on Hugging Face
Parameters3.84B
Context128K
Layers32
Architecturephi3
Smallest quant0.6 GB
Quick answer

phi4 multimodal quantisized needs about 4.1 GB of memory at Q4_K_M (a 2.5 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 2 quants, from 0.6 GB to 2.5 GB (Q4_K_M). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.

Phi‑4 multimodal quantisized – 3.84B parameters, 128K context, vision and audio support

Phi‑4 multimodal quantisized is a 3.84B‑parameter model based on microsoft/Phi‑4‑multimodal‑instruct. It supports text, image, and audio inputs with a 131072‑token context length. The model is available in GGUF format with Q4_K_M quantization.

Good at

  • Multimodal reasoning with text, image, and audio
  • Tool and function calling
  • Long‑context handling

Limits

  • No training from scratch
  • Requires specialized llama.cpp fork
  • License is other, not open source

Best for

  • Local inference on multimodal tasks
  • Research on multimodal reasoning
  • Prototyping agents that consume text, images, and audio

Who it’s for. Developers and researchers who need a lightweight, multimodal LLM for local inference. The model is suitable for applications requiring vision, audio, and text processing within a single framework.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
—0.6 GB 2.2 GB
Fits Get file →
Q4_K_M2.5 GB 4.1 GB
Fits Get file →

For 8 GB VRAM, use the Q4_K_M quantized file (2.49 GB) which needs about 4.1 GB of VRAM at 8 k context. If more VRAM is available, the unquantized 0.61 GB file uses about 2.2 GB of VRAM. Memory estimates are approximate.

What GPU do you need for phi4 multimodal quantisized?

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: the best quant that fits is the main file (0.6 GB file, about 2.2 GB in use).
  • 6 GB VRAM: the best quant that fits is Q4_K_M (2.5 GB file, about 4.1 GB in use).

Which phi4 multimodal quantisized quant should you choose?

Q4_K_M is the usual starting point for phi4 multimodal quantisized: it balances file size and quality well. Here it is a 2.5 GB file.

If memory is tight, the smallest file is the first one in the table at 0.6 GB, with the quality trade-offs that come with fewer bits.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturephi3
Parameters3.84B
Layers32
Trained context131,072 tokens
KV cache (fp16)128 KB per token

Source

Quantized byShayanCyan
FamilyPhi
Licenseother
Downloads (HF)811
Last updated2026-02-16

Frequently asked questions about phi4 multimodal quantisized

How much VRAM does phi4 multimodal quantisized need?

About 4.1 GB at Q4_K_M with an 8K context window. That is the 2.5 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 multimodal quantisized on an 8 GB GPU?

Yes, with Q4_K_M: the 2.5 GB file needs about 4.1 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of phi4 multimodal quantisized GGUF?

It depends on the quant: file is 0.6 GB, Q4_K_M is 2.5 GB.

What context length does phi4 multimodal quantisized support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.3 GB at 32K.

Where can I download phi4 multimodal quantisized GGUF?

The files are on Hugging Face in the ShayanCyan/phi4-multimodal-quantisized-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 multimodal quantisized 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.

Can I run Phi‑4 multimodal on 8 GB VRAM?

Yes, the Q4_K_M quantized file (2.49 GB) requires about 4.1 GB of VRAM at 8 k context, fitting comfortably on 8 GB.

Does this model support vision?

Yes, it includes a multimodal projector for vision and audio, enabling image and audio input processing.

Is the model open source?

The model is released under an other license; it is not open source and cannot be used for training from scratch.

Other GGUFs of Phi-4-multimodal-instruct

Similar models

Browse all →

More Phi models

All Phi →

Other small models

All small models →