Gliese Qwen3.5 4B Abliterated Caption needs about 10.0 GB of memory at BF16 (a 8.4 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 4 quants, from 4.5 GB (Q8_0) to 16.8 GB (F32). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.
Gliese Qwen3.5 4B Abliterated Caption – a vision‑language model for detailed image captioning
Gliese Qwen3.5 4B Abliterated Caption is a 4.21B parameter model built on Qwen3.5. It is optimized for unfiltered, highly detailed image captioning with advanced refusal direction analysis. The model supports vision, reasoning, and tools.
Good at
- Detailed image captioning
- Vision‑language reasoning
- Low‑refusal generation
Limits
- No text‑only inference without vision
- Requires 8k context for full performance
- Apache‑2.0 license limits commercial use without compliance
Best for
- Research on image captioning
- Developing captioning tools
- Testing vision‑language models
Who it’s for. Researchers and developers needing a high‑detail captioning model. The model is intended for research and learning purposes only; users must ensure safe, ethical use.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q8_0 | 4.5 GB | 6.1 GB | Fits | Get file → | |
| BF16 | 8.4 GB | 10.0 GB | Too big | Get file → | |
| F16 | 8.4 GB | 10.0 GB | Too big | Get file → | |
| F32 | 16.8 GB | 18.4 GB | Too big | Get file → |
For 8GB VRAM, use Q8_0 (file size 4.48GB, estimated 6.1GB memory at 8k context). If more VRAM is available, BF16 or F16 (8.42GB, ~10GB memory) can be used. F32 (16.83GB, ~18.4GB memory) requires large memory budgets.
What GPU do you need for Gliese Qwen3.5 4B Abliterated Caption?
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).
- 16 GB VRAM: the best quant that fits is BF16 (8.4 GB file, about 10.0 GB in use).
- 24 GB VRAM: the best quant that fits is F32 (16.8 GB file, about 18.4 GB in use).
Which Gliese Qwen3.5 4B Abliterated Caption quant should you choose?
BF16 (8.4 GB) is the middle option in this repository. The unquantized model. It gives original quality and needs the most memory, so it is mainly for conversion or reference.
If memory is tight, the smallest file is Q8_0 at 4.5 GB, with the quality trade-offs that come with fewer bits. 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
Source
Frequently asked questions about Gliese Qwen3.5 4B Abliterated Caption
How much VRAM does Gliese Qwen3.5 4B Abliterated Caption need?
About 10.0 GB at BF16 with an 8K context window. That is the 8.4 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 Gliese Qwen3.5 4B Abliterated Caption 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 Gliese Qwen3.5 4B Abliterated Caption GGUF?
It depends on the quant: Q8_0 is 4.5 GB, BF16 is 8.4 GB, F16 is 8.4 GB, F32 is 16.8 GB.
What context length does Gliese Qwen3.5 4B Abliterated Caption 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 Gliese Qwen3.5 4B Abliterated Caption GGUF?
The files are on Hugging Face in the prithivMLmods/Gliese-Qwen3.5-4B-Abliterated-Caption repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Gliese Qwen3.5 4B Abliterated Caption 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 8GB VRAM?
Yes, the Q8_0 quant uses about 6.1GB memory at 8k context, fitting within 8GB VRAM.
Does the model support vision?
Yes, it is a vision‑language model with a vision encoder for image captioning.
What license does the model use?
The model is released under the Apache‑2.0 license.