MakeGemma3 Abliterated Improved needs about 22.0 GB of memory at Q4_K_M (a 17.3 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 24 quants, from 0.0 GB to 23.3 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 16.6 GB.
MakeGemma3 Abliterated Improved – 28.4B gemma3 language model
MakeGemma3 Abliterated Improved is a 28.4B parameter gemma3 model merged with NuSLERP. It supports 131072 context tokens and integrates multilingual and medical knowledge. The model is available in multiple GGUF quantizations.
Good at
- Multilingual text generation
- Medical knowledge queries
- Large context handling
Limits
- No vision support in the base GGUF
- License not specified
- Requires 8k context for memory estimates
Best for
- Chatbots needing deep context
- Research assistants for medical literature
- Content creation with multilingual support
Who it’s for. Developers building LLM applications that need a large, multilingual, and medically informed language model. The model is suitable for use with llama.cpp and other GGUF-compatible runtimes.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.0 GB | 4.7 GB | Fits | Get file → | |
| IQ1_S | 6.7 GB | 11.4 GB | Too big | Get file → | |
| IQ1_M | 7.3 GB | 11.9 GB | Too big | Get file → | |
| IQ2_XXS | 8.1 GB | 12.8 GB | Too big | Get file → | |
| IQ2_XS | 8.9 GB | 13.6 GB | Too big | Get file → | |
| IQ2_S | 9.4 GB | 14.0 GB | Too big | Get file → | |
| IQ2_M | 10.1 GB | 14.8 GB | Too big | Get file → | |
| Q2_K_S | 10.2 GB | 14.9 GB | Too big | Get file → | |
| Q2_K | 11.0 GB | 15.6 GB | Too big | Get file → | |
| IQ3_XXS | 11.3 GB | 16.0 GB | Too big | Get file → | |
| IQ3_XS | 12.2 GB | 16.8 GB | Too big | Get file → | |
| IQ3_S | 12.8 GB | 17.4 GB | Too big | Get file → | |
| Q3_K_S | 12.8 GB | 17.4 GB | Too big | Get file → | |
| IQ3_M | 13.2 GB | 17.8 GB | Too big | Get file → | |
| Q3_K_M | 14.0 GB | 18.7 GB | Too big | Get file → | |
| Q3_K_L | 15.1 GB | 19.8 GB | Too big | Get file → | |
| IQ4_XS | 15.5 GB | 20.2 GB | Too big | Get file → | |
| Q4_0 | 16.4 GB | 21.1 GB | Too big | Get file → | |
| Q4_K_S | 16.5 GB | 21.1 GB | Too big | Get file → | |
| Q4_K_M | 17.3 GB | 22.0 GB | Too big | Get file → | |
| Q4_1 | 18.0 GB | 22.7 GB | Too big | Get file → | |
| Q5_K_S | 19.7 GB | 24.4 GB | Too big | Get file → | |
| Q5_K_M | 20.2 GB | 24.9 GB | Too big | Get file → | |
| Q6_K | 23.3 GB | 28.0 GB | Too big | Get file → |
For 8 GB VRAM, use IQ1_S (file size 6.73 GB, estimated 11.4 GB memory at 8k context). For 16 GB VRAM, Q4_K_M (file size 17.34 GB, estimated 22 GB memory) or Q5_K_M (20.24 GB, 24.9 GB memory). For 32 GB VRAM, Q6_K (23.32 GB, 28 GB memory) provides the highest quality.
What GPU do you need for MakeGemma3 Abliterated Improved?
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: the best quant that fits is the main file (0.0 GB file, about 4.7 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 4.7 GB in use).
- 12 GB VRAM: the best quant that fits is IQ1_M (7.3 GB file, about 11.9 GB in use).
- 16 GB VRAM: the best quant that fits is IQ3_XXS (11.3 GB file, about 16.0 GB in use).
- 24 GB VRAM: the best quant that fits is Q4_1 (18.0 GB file, about 22.7 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K (23.3 GB file, about 28.0 GB in use).
Which MakeGemma3 Abliterated Improved quant should you choose?
Q4_K_M is the usual starting point for MakeGemma3 Abliterated Improved: it balances file size and quality well. Here it is a 17.3 GB file.
If memory is tight, the smallest file is the first one in the table at 0.0 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
Frequently asked questions about MakeGemma3 Abliterated Improved
How much VRAM does MakeGemma3 Abliterated Improved need?
About 22.0 GB at Q4_K_M with an 8K context window. That is the 17.3 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 MakeGemma3 Abliterated Improved on an 8 GB GPU?
Yes, with the smaller files: the 0.0 GB file needs about 4.7 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of MakeGemma3 Abliterated Improved GGUF?
It depends on the quant: file is 0.0 GB, IQ1_S is 6.7 GB, IQ1_M is 7.3 GB, IQ2_XXS is 8.1 GB, IQ2_XS is 8.9 GB, IQ2_S is 9.4 GB and 18 more.
What context length does MakeGemma3 Abliterated Improved support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 4.2 GB at 8K and 16.6 GB at 32K.
Where can I download MakeGemma3 Abliterated Improved GGUF?
The files are on Hugging Face in the mradermacher/MakeGemma3-abliterated-Improved-i1-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
Can I run this model on 8 GB VRAM?
The IQ1_S quant requires about 11.4 GB memory at 8k context, so 8 GB VRAM is insufficient. Use a smaller quant or increase VRAM.
Does this model support vision tasks?
The base GGUF does not include vision support. Vision functionality requires an mmproj file from the static repository.
What is the context length of this model?
The model supports 131072 tokens of context.