Run unsloth/gemma-4-E4B-it-GGUF locally
unsloth/gemma-4-E4B-it-GGUF is a mid-size instruction-tuned chat model with 7.52 billion parameters, built on the gemma4 architecture. It is released under the apache-2.0 license and has been downloaded 687,378 times.
To run unsloth/gemma-4-E4B-it-GGUF locally at a 4,096-token context, its quantized versions need between 0.9 GB (GGUF, lowest quality) and 15.91 GB (BF16, highest quality) of memory, weights plus KV cache and a system margin included.
For most users the best balance is Q6_K_XL, needing about 7.75 GB. That means unsloth/gemma-4-E4B-it-GGUF fits entirely in the VRAM of a 6 GB GPU or larger, running fully on the GPU.
All quantizations
| Quant. | Bits | Quality | Weights | KV | Total | Speed~ | Verdict |
|---|---|---|---|---|---|---|---|
| GGUF | 0.1 | Very low | 0.09 GB | 0.01 GB | 0.9 GB | 4353.6 t/s | Fits in VRAM |
| F16 | 1.24 | Very low | 1.08 GB | 0.01 GB | 1.89 GB | 369.6 t/s | Fits in VRAM |
| F32 | 2.04 | Very low | 1.78 GB | 0.01 GB | 2.59 GB | 224.6 t/s | Fits in VRAM |
| IQ2_M | 3.77 | Fair | 3.3 GB | 0.01 GB | 4.11 GB | 121.2 t/s | Fits in VRAM |
| IQ3_XXS | 3.96 | Fair | 3.46 GB | 0.01 GB | 4.27 GB | 115.5 t/s | Fits in VRAM |
| Q2_K_XL | 4.0 | Fair | 3.5 GB | 0.01 GB | 4.31 GB | 114.3 t/s | Fits in VRAM |
| Q3_K_S | 4.11 | Fair | 3.6 GB | 0.01 GB | 4.4 GB | 111.2 t/s | Fits in VRAM |
| Q3_K_M | 4.32 | Good | 3.78 GB | 0.01 GB | 4.59 GB | 105.8 t/s | Fits in VRAM |
| Q3_K_XL | 4.88 | Good | 4.27 GB | 0.01 GB | 5.08 GB | 93.6 t/s | Fits in VRAM |
| IQ4_XS | 5.02 | Very good | 4.39 GB | 0.01 GB | 5.2 GB | 91.1 t/s | Fits in VRAM |
| IQ4_NL | 5.15 | Very good | 4.5 GB | 0.01 GB | 5.31 GB | 88.8 t/s | Fits in VRAM |
| Q4_0 | 5.15 | Very good | 4.5 GB | 0.01 GB | 5.31 GB | 88.8 t/s | Fits in VRAM |
| Q4_K_S | 5.16 | Very good | 4.51 GB | 0.01 GB | 5.32 GB | 88.7 t/s | Fits in VRAM |
| Q4_K_M | 5.3 | Very good | 4.64 GB | 0.01 GB | 5.44 GB | 86.3 t/s | Fits in VRAM |
| Q4_1 | 5.4 | Very good | 4.73 GB | 0.01 GB | 5.53 GB | 84.6 t/s | Fits in VRAM |
| Q4_K_XL | 5.45 | Very good | 4.77 GB | 0.01 GB | 5.58 GB | 83.8 t/s | Fits in VRAM |
| Q5_K_S | 5.75 | Very good | 5.03 GB | 0.01 GB | 5.84 GB | 79.5 t/s | Fits in VRAM |
| Q5_K_M | 5.83 | Very good | 5.11 GB | 0.01 GB | 5.91 GB | 78.3 t/s | Fits in VRAM |
| Q5_K_XL | 7.08 | Excellent | 6.2 GB | 0.01 GB | 7.01 GB | 64.5 t/s | Fits in VRAM |
| Q6_K | 7.53 | Excellent | 6.59 GB | 0.01 GB | 7.4 GB | 60.7 t/s | Fits in VRAM |
| Q6_K_XL | 7.94 | Excellent | 6.95 GB | 0.01 GB | 7.75 GB | 57.6 t/s | Fits in VRAM |
| Q8_0 | 8.82 | Excellent | 7.72 GB | 0.01 GB | 8.53 GB | 6.5 t/s | Offload |
| Q8_K_XL | 9.27 | Excellent | 8.11 GB | 0.01 GB | 8.92 GB | 6.2 t/s | Offload |
| BF16 | 17.26 | Excellent | 15.1 GB | 0.01 GB | 15.91 GB | 3.3 t/s | Offload |
KV cache computed from the model's exact architecture. Speed is a rough estimate bounded by memory bandwidth.
Frequently asked questions
How much VRAM do you need to run unsloth/gemma-4-E4B-it-GGUF?
You need about 5.91 GB of VRAM to run unsloth/gemma-4-E4B-it-GGUF entirely on the GPU using the Q5_K_M quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.
Can I run unsloth/gemma-4-E4B-it-GGUF on an 8 GB GPU?
Yes. With 8 GB of VRAM you can run unsloth/gemma-4-E4B-it-GGUF fully on the GPU using Q6_K_XL (about 7.75 GB).
Can I run unsloth/gemma-4-E4B-it-GGUF on a 16 GB GPU?
Yes. With 16 GB of VRAM you can run unsloth/gemma-4-E4B-it-GGUF fully on the GPU using BF16 (about 15.91 GB).
Can I run unsloth/gemma-4-E4B-it-GGUF on a 24 GB GPU?
Yes. With 24 GB of VRAM you can run unsloth/gemma-4-E4B-it-GGUF fully on the GPU using BF16 (about 15.91 GB).
What is the best quantization for unsloth/gemma-4-E4B-it-GGUF?
If memory allows, higher bits-per-weight means better quality. A common sweet spot is a Q4_K_M or Q5_K_M quantization, which keeps most of the quality while roughly halving the memory versus 8-bit. Pick the highest quantization that still fits in your VRAM.