GE

gemma3

GemmaGGUF by cortexsoUpdated 2025-05-13
Download on Hugging Face
Parameters11.8B
Context32K
Layers26
Architecturegemma3
Smallest quant0.7 GB
Quick answer

gemma3 needs about 1.5 GB of memory at Q4_K_M (a 0.8 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 40 quants, from 0.7 GB (Q3_K_S) to 28.7 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 0.9 GB.

Gemma 3 11.8B parameter model with 32K context

Gemma 3 is a 11.8B parameter language model from the Gemma family. It supports a 32K token context window and multimodal input. The model is available in several quantized GGUF variants.

Good at

  • Large context handling
  • Multimodal text and image input
  • Efficient inference

Limits

  • No vision support in this text‑only variant
  • Requires 32K context for full capability
  • License is Gemma

Best for

  • Chat applications needing long context
  • Text summarization with extended context
  • Instruction‑tuned tasks

Who it’s for. Developers building LLM applications on laptops or desktops with limited GPU memory. The model is open‑weight and can be run with llama.cpp or Cortex CLI.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q3_K_S0.7 GB 1.4 GB
Fits Get file →
Q2_K0.7 GB 1.4 GB
Fits Get file →
Q3_K_M0.7 GB 1.4 GB
Fits Get file →
Q3_K_L0.8 GB 1.5 GB
Fits Get file →
Q4_K_S0.8 GB 1.5 GB
Fits Get file →
Q4_K_M0.8 GB 1.5 GB
Fits Get file →
Q5_K_S0.8 GB 1.6 GB
Fits Get file →
Q5_K_M0.9 GB 1.6 GB
Fits Get file →
Q6_K1.0 GB 1.7 GB
Fits Get file →
Q8_01.1 GB 1.8 GB
Fits Get file →
Q2_K1.7 GB 2.4 GB
Fits Get file →
Q3_K_S1.9 GB 2.7 GB
Fits Get file →
Q3_K_M2.1 GB 2.8 GB
Fits Get file →
Q3_K_L2.2 GB 3.0 GB
Fits Get file →
Q4_K_S2.4 GB 3.1 GB
Fits Get file →
Q4_K_M2.5 GB 3.2 GB
Fits Get file →
Q5_K_S2.8 GB 3.5 GB
Fits Get file →
Q5_K_M2.8 GB 3.5 GB
Fits Get file →
Q6_K3.2 GB 3.9 GB
Fits Get file →
Q8_04.1 GB 4.8 GB
Fits Get file →
Q2_K4.8 GB 5.5 GB
Fits Get file →
Q3_K_S5.5 GB 6.2 GB
Fits Get file →
Q3_K_M6.0 GB 6.7 GB
Fits Get file →
Q3_K_L6.5 GB 7.2 GB
Fits Get file →
Q4_K_S6.9 GB 7.7 GB
Tight Get file →
Q4_K_M7.3 GB 8.0 GB
Too big Get file →
Q5_K_S8.2 GB 8.9 GB
Too big Get file →
Q5_K_M8.4 GB 9.2 GB
Too big Get file →
Q6_K9.7 GB 10.4 GB
Too big Get file →
Q2_K10.5 GB 11.2 GB
Too big Get file →
Q3_K_S12.2 GB 12.9 GB
Too big Get file →
Q8_012.5 GB 13.2 GB
Too big Get file →
Q3_K_M13.4 GB 14.2 GB
Too big Get file →
Q3_K_L14.5 GB 15.3 GB
Too big Get file →
Q4_K_S15.7 GB 16.4 GB
Too big Get file →
Q4_K_M16.5 GB 17.3 GB
Too big Get file →
Q5_K_S18.8 GB 19.5 GB
Too big Get file →
Q5_K_M19.3 GB 20.0 GB
Too big Get file →
Q6_K22.2 GB 22.9 GB
Too big Get file →
Q8_028.7 GB 29.4 GB
Too big Get file →

For 8 GB VRAM, use Q4_K_M (file size 0.81 GB, memory 1.5 GB). For 16 GB VRAM, Q5_K_M (file size 2.83 GB, memory 3.5 GB) is suitable. For 32 GB VRAM, Q6_K (file size 3.19 GB, memory 3.9 GB) offers higher precision.

What GPU do you need for gemma3?

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 Q6_K (3.2 GB file, about 3.9 GB in use).
  • 6 GB VRAM: the best quant that fits is Q2_K (4.8 GB file, about 5.5 GB in use).
  • 8 GB VRAM: the best quant that fits is Q4_K_M (7.3 GB file, about 8.0 GB in use).
  • 12 GB VRAM: the best quant that fits is Q2_K (10.5 GB file, about 11.2 GB in use).
  • 16 GB VRAM: the best quant that fits is Q3_K_L (14.5 GB file, about 15.3 GB in use).
  • 24 GB VRAM: the best quant that fits is Q6_K (22.2 GB file, about 22.9 GB in use).
  • 32 GB VRAM: the best quant that fits is Q8_0 (28.7 GB file, about 29.4 GB in use).

Which gemma3 quant should you choose?

Q4_K_M is the usual starting point for gemma3: it balances file size and quality well. Here it is a 0.8 GB file.

The Q8_0 file is 1.1 GB, about 1.3 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

If memory is tight, the smallest file is Q3_K_S at 0.7 GB, with the quality trade-offs that come with fewer bits. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturegemma3
Parameters11.8B
Layers26
Trained context32,768 tokens
KV cache (fp16)26 KB per token

Source

Quantized bycortexso
FamilyGemma
Repositorycortexso/gemma3
Licensegemma
Downloads (HF)570
Last updated2025-05-13

Frequently asked questions about gemma3

How much VRAM does gemma3 need?

About 1.5 GB at Q4_K_M with an 8K context window. That is the 0.8 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 gemma3 on an 8 GB GPU?

Yes, with Q4_K_M: the 7.3 GB file needs about 8.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of gemma3 GGUF?

It depends on the quant: Q3_K_S is 0.7 GB, Q2_K is 0.7 GB, Q3_K_M is 0.7 GB, Q3_K_L is 0.8 GB, Q4_K_S is 0.8 GB, Q4_K_M is 0.8 GB and 34 more.

What context length does gemma3 support?

The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 0.2 GB at 8K and 0.9 GB at 32K.

Where can I download gemma3 GGUF?

The files are on Hugging Face in the cortexso/gemma3 repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does gemma3 use?

Hugging Face lists the license as gemma. 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 Gemma 3 on 8 GB VRAM?

Yes, a Q4_K_M quantized file uses about 1.5 GB of VRAM at 8K context.

Does Gemma 3 support image input?

The base model supports multimodal input, but the provided GGUF files are text‑only variants.

Similar models

Browse all →

More from cortexso

All by cortexso →

More Gemma models

All Gemma →

Other medium models

All medium models →