gemma-3-270m-it GGUF size and VRAM requirements

License: gemma ⬇ 89,019 ❤ 171
Parameters0.27B
Context32,768

unsloth/gemma-3-270m-it-GGUF is a compact instruction-tuned chat model with 0.27 billion parameters, built on the gemma3 architecture. It is released under the gemma license and has been downloaded 89,019 times.

Under the hood it uses 18 transformer layers, a hidden size of 640, 4 attention heads. It uses grouped-query attention (4 query heads sharing 1 key/value heads), which already trims KV-cache memory compared with full multi-head attention.

To run unsloth/gemma-3-270m-it-GGUF locally at a 4,096-token context, its quantized versions need between 1.01 GB (IQ2_XXS, lowest quality) and 1.35 GB (F16, highest quality) of memory, weights plus KV cache and a system margin included. Context length drives that memory directly: at 4,096 tokens the KV cache for gemma-3-270m-it-GGUF is about 0.04 GB, rising to roughly 0.35 GB at its full 32,768-token context. Shorter prompts free up memory for a higher-quality quantization.

For most users the best balance is F16, needing about 1.35 GB. That means unsloth/gemma-3-270m-it-GGUF fits entirely in the VRAM of a 6 GB GPU or larger, running fully on the GPU.

Available GGUF quantizations for unsloth/gemma-3-270m-it-GGUF include IQ2_XXS, IQ2_M, IQ3_XXS, Q3_K_S, Q2_K, Q2_K_L, Q2_K_XL, IQ4_XS, Q4_0, IQ4_NL, Q3_K_M, Q3_K_XL, Q4_1, Q4_K_S, Q4_K_M, Q4_K_XL, Q5_K_S, Q5_K_M, Q5_K_XL, Q6_K, Q6_K_XL, Q8_0, Q8_K_XL, F16. The model supports a native context length of up to 32,768 tokens; a longer context grows the KV cache and the memory needed.

→ Guide: How much VRAM do you need?

GGUF file size and memory by quantization

Compare real GGUF weight sizes, estimated KV cache and total memory for Q4, Q5, Q8 and every quantization published in this repository.

Quant.Bits QualityWeights KVTotal Speed~Verdict
IQ2_XXS 5.37 Very good 0.17 GB 0.04 GB 1.01 GB 2384.7 t/s Fits in VRAM
IQ2_M 5.45 Very good 0.17 GB 0.04 GB 1.01 GB 2349.7 t/s Fits in VRAM
IQ3_XXS 5.51 Very good 0.17 GB 0.04 GB 1.02 GB 2327.7 t/s Fits in VRAM
Q3_K_S 7.06 Excellent 0.22 GB 0.04 GB 1.06 GB 1814.4 t/s Fits in VRAM
Q2_K 7.07 Excellent 0.22 GB 0.04 GB 1.06 GB 1811.6 t/s Fits in VRAM
Q2_K_L 7.07 Excellent 0.22 GB 0.04 GB 1.06 GB 1811.6 t/s Fits in VRAM
Q2_K_XL 7.09 Excellent 0.22 GB 0.04 GB 1.07 GB 1806.5 t/s Fits in VRAM
IQ4_XS 7.19 Excellent 0.22 GB 0.04 GB 1.07 GB 1783.2 t/s Fits in VRAM
Q4_0 7.21 Excellent 0.22 GB 0.04 GB 1.07 GB 1777.9 t/s Fits in VRAM
IQ4_NL 7.22 Excellent 0.23 GB 0.04 GB 1.07 GB 1775.0 t/s Fits in VRAM
Q3_K_M 7.22 Excellent 0.23 GB 0.04 GB 1.07 GB 1775.0 t/s Fits in VRAM
Q3_K_XL 7.24 Excellent 0.23 GB 0.04 GB 1.07 GB 1770.2 t/s Fits in VRAM
Q4_1 7.39 Excellent 0.23 GB 0.04 GB 1.07 GB 1734.1 t/s Fits in VRAM
Q4_K_S 7.46 Excellent 0.23 GB 0.04 GB 1.08 GB 1718.7 t/s Fits in VRAM
Q4_K_M 7.55 Excellent 0.24 GB 0.04 GB 1.08 GB 1696.8 t/s Fits in VRAM
Q4_K_XL 7.58 Excellent 0.24 GB 0.04 GB 1.08 GB 1691.4 t/s Fits in VRAM
Q5_K_S 7.7 Excellent 0.24 GB 0.04 GB 1.08 GB 1664.7 t/s Fits in VRAM
Q5_K_M 7.76 Excellent 0.24 GB 0.04 GB 1.09 GB 1651.7 t/s Fits in VRAM
Q5_K_XL 7.76 Excellent 0.24 GB 0.04 GB 1.09 GB 1651.7 t/s Fits in VRAM
Q6_K 8.44 Excellent 0.26 GB 0.04 GB 1.11 GB 1517.8 t/s Fits in VRAM
Q6_K_XL 8.54 Excellent 0.27 GB 0.04 GB 1.11 GB 1501.0 t/s Fits in VRAM
Q8_0 8.7 Excellent 0.27 GB 0.04 GB 1.12 GB 1473.2 t/s Fits in VRAM
Q8_K_XL 14.06 Excellent 0.44 GB 0.04 GB 1.28 GB 911.7 t/s Fits in VRAM
F16 16.2 Excellent 0.51 GB 0.04 GB 1.35 GB 791.2 t/s Fits in VRAM

KV cache computed from the model's exact architecture. Speed is a rough estimate bounded by memory bandwidth.

Frequently asked questions

What kind of model is unsloth/gemma-3-270m-it-GGUF?

unsloth/gemma-3-270m-it-GGUF is an instruction-tuned chat model with 0.27 billion parameters, based on the gemma3 architecture. It is released under the gemma license and distributed as GGUF files for local inference.

How much VRAM do you need to run unsloth/gemma-3-270m-it-GGUF?

You need about 1.35 GB of VRAM to run unsloth/gemma-3-270m-it-GGUF entirely on the GPU using the F16 quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.

Can I run unsloth/gemma-3-270m-it-GGUF on an 8 GB GPU?

Yes. With 8 GB of VRAM you can run unsloth/gemma-3-270m-it-GGUF fully on the GPU using F16 (about 1.35 GB).

Can I run unsloth/gemma-3-270m-it-GGUF on a 16 GB GPU?

Yes. With 16 GB of VRAM you can run unsloth/gemma-3-270m-it-GGUF fully on the GPU using F16 (about 1.35 GB).

Can I run unsloth/gemma-3-270m-it-GGUF on a 24 GB GPU?

Yes. With 24 GB of VRAM you can run unsloth/gemma-3-270m-it-GGUF fully on the GPU using F16 (about 1.35 GB).

What context length does unsloth/gemma-3-270m-it-GGUF support?

unsloth/gemma-3-270m-it-GGUF supports a native context length of up to 32,768 tokens. A longer context grows the KV cache, so it increases the memory needed to run the model.

What is the best quantization for unsloth/gemma-3-270m-it-GGUF?

For unsloth/gemma-3-270m-it-GGUF, a strong default is Q4_K_M, which needs about 1.08 GB and keeps most of the quality while roughly halving the memory versus 8-bit. With VRAM to spare, Q5_K_M or Q6_K add a little more quality; if you are tight on memory, a smaller quantization still runs. Pick the highest quantization that fits your VRAM.