QW

qwen3 vl 2b instruct q4 k m ggml llama cpp vivo x300

QwenGGUF by ArmUpdated 2026-10-13
Download on Hugging Face
Parameters1.72B
Context256K
Layers28
Architectureqwen3vl
Smallest quant1.1 GB
Quick answer

qwen3 vl 2b instruct q4 k m ggml llama cpp vivo x300 needs about 2.5 GB of memory at its main file (a 1.1 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 2 quants, from 1.1 GB to 3.4 GB. Longer context needs more: at 32K tokens the KV cache alone adds about 3.8 GB.

Qwen3‑VL‑2B‑Instruct optimized for Arm mobile CPUs

Qwen3‑VL‑2B‑Instruct is a 1.72 billion‑parameter vision‑language model. It supports tools and vision tasks and is quantized for llama.cpp on Arm CPUs with SME2. The model offers a 262144‑token context and is released under Apache‑2.0.

Good at

  • Vision‑language generation
  • Tool invocation
  • Spatial reasoning

Limits

  • No vision support without paired mmproj file
  • Requires 8 k context for memory estimates
  • No GPU acceleration on non‑Arm CPUs

Best for

  • Mobile image captioning
  • Interactive visual agents
  • Embedded multimodal assistants

Who it’s for. Developers building mobile or embedded applications that need vision‑language capabilities on Arm CPUs. The model is suitable for edge devices where memory is limited but inference speed is critical.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
—1.1 GB 2.5 GB
Fits Get file →
—3.4 GB 4.9 GB
Fits Get file →

For 8 GB VRAM, use the 1.11 GB file (est. 2.5 GB memory at 8 k context). For 16 GB VRAM, the 3.45 GB file (est. 4.9 GB memory) provides higher quality. Memory estimates are approximate.

What GPU do you need for qwen3 vl 2b instruct q4 k m ggml llama cpp vivo x300?

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 (1.1 GB file, about 2.5 GB in use).
  • 6 GB VRAM: the best quant that fits is the main file (3.4 GB file, about 4.9 GB in use).

Specifications

Architecture from the GGUF header

Architectureqwen3vl
Parameters1.72B
Layers28
Trained context262,144 tokens
KV cache (fp16)112 KB per token

Source

Quantized byArm
FamilyQwen
Licenseapache-2.0
Downloads (HF)770
Last updated2026-10-13

Frequently asked questions about qwen3 vl 2b instruct q4 k m ggml llama cpp vivo x300

How much VRAM does qwen3 vl 2b instruct q4 k m ggml llama cpp vivo x300 need?

About 2.5 GB at its main file with an 8K context window. That is the 1.1 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 qwen3 vl 2b instruct q4 k m ggml llama cpp vivo x300 on an 8 GB GPU?

Yes, with the smaller files: the 3.4 GB file needs about 4.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of qwen3 vl 2b instruct q4 k m ggml llama cpp vivo x300 GGUF?

It depends on the quant: file is 1.1 GB, file is 3.4 GB.

What context length does qwen3 vl 2b instruct q4 k m ggml llama cpp vivo x300 support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.9 GB at 8K and 3.8 GB at 32K.

Where can I download qwen3 vl 2b instruct q4 k m ggml llama cpp vivo x300 GGUF?

The files are on Hugging Face in the Arm/qwen3-vl-2b-instruct-q4-k-m-ggml-llama-cpp-vivo-x300 repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does qwen3 vl 2b instruct q4 k m ggml llama cpp vivo x300 use?

Hugging Face lists the license as apache-2.0. 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 this model on 8 GB VRAM?

Yes, the 1.11 GB GGUF file requires about 2.5 GB of VRAM at 8 k context, which fits within 8 GB.

Do I need a separate mmproj file?

Yes, the vision encoder and projector are in a separate mmproj file kept at Q8_0. Both files must be loaded together.

Is the model compatible with non‑Arm CPUs?

The quantization is optimized for Arm CPUs with SME2; performance on other CPUs may be lower.

Other GGUFs of Qwen3-VL-2B-Instruct

Similar models

Browse all →

More from Arm

All by Arm →

More Qwen models

All Qwen →

Other tiny models

All tiny models →