Run unsloth/Qwen3.5-9B-GGUF locally

License: apache-2.0 ⬇ 1,047,718 ❤ 727
Parameters8.95B
Context262,144

unsloth/Qwen3.5-9B-GGUF is a large language model with 8.95 billion parameters, built on the qwen35 architecture. It is released under the apache-2.0 license and has been downloaded 1,047,718 times.

To run unsloth/Qwen3.5-9B-GGUF locally at a 4,096-token context, its quantized versions need between 2.13 GB (F16, lowest quality) and 18.82 GB (BF16, highest quality) of memory, weights plus KV cache and a system margin included.

For most users the best balance is Q5_K_XL, needing about 7.56 GB. That means unsloth/Qwen3.5-9B-GGUF fits entirely in the VRAM of a 6 GB GPU or larger, running fully on the GPU.

→ Guide: How much VRAM do you need?

All quantizations

Quant.Bits QualityWeights KVTotal Speed~Verdict
F16 0.82 Very low 0.86 GB 0.47 GB 2.13 GB 467.8 t/s Fits in VRAM
F32 1.63 Very low 1.7 GB 0.47 GB 2.97 GB 235.5 t/s Fits in VRAM
IQ2_XXS 2.85 Low 2.97 GB 0.47 GB 4.25 GB 134.6 t/s Fits in VRAM
IQ2_M 3.26 Low 3.4 GB 0.47 GB 4.67 GB 117.7 t/s Fits in VRAM
IQ3_XXS 3.59 Fair 3.74 GB 0.47 GB 5.02 GB 106.9 t/s Fits in VRAM
Q2_K_XL 3.68 Fair 3.84 GB 0.47 GB 5.11 GB 104.2 t/s Fits in VRAM
Q3_K_S 3.86 Fair 4.02 GB 0.47 GB 5.3 GB 99.5 t/s Fits in VRAM
Q3_K_M 4.18 Fair 4.35 GB 0.47 GB 5.63 GB 91.9 t/s Fits in VRAM
Q3_K_XL 4.52 Good 4.71 GB 0.47 GB 5.98 GB 85.0 t/s Fits in VRAM
IQ4_XS 4.62 Good 4.81 GB 0.47 GB 6.09 GB 83.1 t/s Fits in VRAM
IQ4_NL 4.8 Good 5.0 GB 0.47 GB 6.28 GB 80.0 t/s Fits in VRAM
Q4_0 4.81 Good 5.01 GB 0.47 GB 6.28 GB 79.8 t/s Fits in VRAM
Q4_K_S 4.82 Good 5.02 GB 0.47 GB 6.3 GB 79.6 t/s Fits in VRAM
Q4_K_M 5.08 Very good 5.29 GB 0.47 GB 6.57 GB 75.6 t/s Fits in VRAM
Q4_1 5.22 Very good 5.44 GB 0.47 GB 6.71 GB 73.6 t/s Fits in VRAM
Q4_K_XL 5.33 Very good 5.56 GB 0.47 GB 6.83 GB 72.0 t/s Fits in VRAM
Q5_K_S 5.68 Very good 5.92 GB 0.47 GB 7.2 GB 67.5 t/s Fits in VRAM
Q5_K_M 5.88 Very good 6.13 GB 0.47 GB 7.4 GB 65.3 t/s Fits in VRAM
Q5_K_XL 6.03 Very good 6.28 GB 0.47 GB 7.56 GB 63.7 t/s Fits in VRAM
Q6_K 6.66 Excellent 6.95 GB 0.47 GB 8.22 GB 7.2 t/s Offload
Q6_K_XL 7.82 Excellent 8.16 GB 0.47 GB 9.43 GB 6.1 t/s Offload
Q8_0 8.51 Excellent 8.87 GB 0.47 GB 10.15 GB 5.6 t/s Offload
Q8_K_XL 11.59 Excellent 12.08 GB 0.47 GB 13.36 GB 4.1 t/s Offload
BF16 16.84 Excellent 17.55 GB 0.47 GB 18.82 GB 2.8 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/Qwen3.5-9B-GGUF?

You need about 5.98 GB of VRAM to run unsloth/Qwen3.5-9B-GGUF entirely on the GPU using the Q3_K_XL quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.

Can I run unsloth/Qwen3.5-9B-GGUF on an 8 GB GPU?

Yes. With 8 GB of VRAM you can run unsloth/Qwen3.5-9B-GGUF fully on the GPU using Q5_K_XL (about 7.56 GB).

Can I run unsloth/Qwen3.5-9B-GGUF on a 16 GB GPU?

Yes. With 16 GB of VRAM you can run unsloth/Qwen3.5-9B-GGUF fully on the GPU using Q8_K_XL (about 13.36 GB).

Can I run unsloth/Qwen3.5-9B-GGUF on a 24 GB GPU?

Yes. With 24 GB of VRAM you can run unsloth/Qwen3.5-9B-GGUF fully on the GPU using BF16 (about 18.82 GB).

What is the best quantization for unsloth/Qwen3.5-9B-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.