QW

Qwen3.5 4B Base ZitGen V1

QwenGGUF by lolzinventorUpdated 2026-04-10
Download on Hugging Face
Parameters4.21B
Context256K
Layers32
Architectureqwen35
Smallest quant4.5 GB
Quick answer

Qwen3.5 4B Base ZitGen V1 needs about 6.1 GB of memory at Q8_0 (a 4.5 GB file) with an 8K context window, so it fits on a 8 GB GPU. This repository has 2 quants, from 4.5 GB (Q8_0) to 8.4 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.

Qwen3.5 4B Base ZitGen V1 – Vision‑enabled language model fine‑tuned for image captioning

Qwen3.5 4B Base ZitGen V1 is a 4.21 billion‑parameter causal language model with vision encoder. It was fine‑tuned on LLM‑generated image‑prompt pairs for optimized Zit prompt generation. The model supports 201 languages and multimodal reasoning.

Good at

  • Vision‑language reasoning
  • Image captioning
  • Prompt generation

Limits

  • No direct vision support in llama.cpp
  • Requires 8 k context for memory estimates
  • License is apache‑2.0

Best for

  • Generating optimized image prompts
  • Fine‑tuning vision‑language tasks
  • Research on multimodal LLMs

Who it’s for. Developers and researchers needing a lightweight vision‑language model for prompt generation and fine‑tuning. The model is suitable for environments with limited GPU memory and for projects that require open‑source licensing.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q8_04.5 GB 6.1 GB
Fits Get file →
BF168.4 GB 10.0 GB
Too big Get file →

For 8 GB VRAM, use Q8_0 (4.48 GB file, ~6.1 GB memory at 8 k context). For 16 GB VRAM, BF16 (8.42 GB file, ~10 GB memory) offers higher precision. Memory estimates are approximate.

What GPU do you need for Qwen3.5 4B Base ZitGen V1?

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: the best quant that fits is Q8_0 (4.5 GB file, about 6.1 GB in use).
  • 12 GB VRAM: the best quant that fits is BF16 (8.4 GB file, about 10.0 GB in use).

Which Qwen3.5 4B Base ZitGen V1 quant should you choose?

Q8_0 (4.5 GB) is the middle option in this repository. Close to lossless. It costs roughly twice the memory of a 4-bit file, so it suits cases where you have plenty of VRAM.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architectureqwen35
Parameters4.21B
Layers32
Trained context262,144 tokens
KV cache (fp16)128 KB per token

Source

Quantized bylolzinventor
FamilyQwen
Licenseapache-2.0
Downloads (HF)1,143
Last updated2026-04-10

Frequently asked questions about Qwen3.5 4B Base ZitGen V1

How much VRAM does Qwen3.5 4B Base ZitGen V1 need?

About 6.1 GB at Q8_0 with an 8K context window. That is the 4.5 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 Qwen3.5 4B Base ZitGen V1 on an 8 GB GPU?

Yes, with Q8_0: the 4.5 GB file needs about 6.1 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Qwen3.5 4B Base ZitGen V1 GGUF?

It depends on the quant: Q8_0 is 4.5 GB, BF16 is 8.4 GB.

What context length does Qwen3.5 4B Base ZitGen V1 support?

The GGUF header lists a trained context length of 262,144 tokens (256K). 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 Qwen3.5 4B Base ZitGen V1 GGUF?

The files are on Hugging Face in the lolzinventor/Qwen3.5-4B-Base-ZitGen-V1 repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Qwen3.5 4B Base ZitGen V1 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 Qwen3.5 4B Base ZitGen V1 on 8 GB VRAM?

Yes, the Q8_0 quantized model is 4.48 GB and requires about 6.1 GB memory at 8 k context, fitting within 8 GB VRAM.

Does the model support vision tasks in llama.cpp?

The model includes a vision encoder but llama.cpp does not natively support vision; external handling is required.

What license applies to this model?

The model is released under the apache‑2.0 license.

Other GGUFs of Qwen3.5-4B-Base

Similar models

Browse all →

More Qwen models

All Qwen →

Other small models

All small models →