Qwen3.5 27B Instruct quants compared

27 real GGUF builds of Qwen3.5 27B Instruct measured from bartowski/Qwen_Qwen3.5-27B-GGUF. Download the Q4_K_M (16.75 GB) unless you know why you need heavier. Probed 2026-09-02.

Download Qwen3.5 27B Instruct Q4_K_M (16.75 GB). It fits 24 GB of memory with 16k context. With Ollama: ollama run qwen3.5:27b

Every Qwen3.5 27B Instruct quant by real file size

QuantWeights+ KV (16k)TotalFits comfortably inQuality
BF1650.9 GB1.0 GB51.9 GB64 GBFull precision (lossless)
Q8_027.12 GB1.0 GB28.1 GB32 GBNear-lossless
Q6_K_L22.62 GB1.0 GB23.6 GB32 GBExcellent
Q6_K22.05 GB1.0 GB23.1 GB32 GBExcellent
Q5_K_L20.26 GB1.0 GB21.3 GB24 GBVery high
Q5_K_M19.53 GB1.0 GB20.5 GB24 GBVery high
Q5_K_S18.53 GB1.0 GB19.5 GB24 GBVery high
Q4_K_M *16.75 GB1.0 GB17.8 GB24 GBHigh: the default pick
Q4_K_L17.63 GB1.0 GB18.6 GB24 GBHigh
Q4_116.8 GB1.0 GB17.8 GB24 GBHigh
Q4_K_S15.76 GB1.0 GB16.8 GB24 GBHigh
Q4_015.42 GB1.0 GB16.4 GB24 GBHigh
IQ4_NL15.4 GB1.0 GB16.4 GB24 GBHigh
IQ4_XS14.7 GB1.0 GB15.7 GB24 GBHigh
Q3_K_XL15.46 GB1.0 GB16.5 GB24 GBAcceptable: visible loss
Q3_K_L14.43 GB1.0 GB15.4 GB24 GBAcceptable: visible loss
Q3_K_M13.8 GB1.0 GB14.8 GB24 GBAcceptable: visible loss
IQ3_M13.15 GB1.0 GB14.2 GB16 GBAcceptable: visible loss
Q3_K_S12.98 GB1.0 GB14.0 GB16 GBAcceptable: visible loss
IQ3_XS12.61 GB1.0 GB13.6 GB16 GBAcceptable: visible loss
IQ3_XXS11.96 GB1.0 GB13.0 GB16 GBAcceptable: visible loss
Q2_K_L12.38 GB1.0 GB13.4 GB16 GBExperimental: not ranked: never recommended
Q2_K11.22 GB1.0 GB12.2 GB16 GBExperimental: not ranked: never recommended
IQ2_M10.32 GB1.0 GB11.3 GB16 GBExperimental: not ranked: never recommended
IQ2_S9.79 GB1.0 GB10.8 GB12 GBExperimental: not ranked: never recommended
IQ2_XS9.5 GB1.0 GB10.5 GB12 GBExperimental: not ranked: never recommended
IQ2_XXS8.95 GB1.0 GB9.9 GB12 GBExperimental: not ranked: never recommended

* default pick. Weights = real GGUF file sizes from bartowski/Qwen_Qwen3.5-27B-GGUF (probed 2026-09-02). KV = fp16 estimate; a q8_0 cache roughly halves it. "Comfortable" = weights + KV within 90% of memory.

Best Qwen3.5 27B Instruct quant by memory

MemoryRecommended quantTotal (16k ctx)
16 GBIQ3_XXS13.0 GB
24 GBQ5_K_M20.5 GB
32 GBQ8_028.1 GB
64 GBBF1651.9 GB

Why we don't rank Qwen3.5 27B Instruct's 2-bit quants

Quants at 2 bits per weight or below (Q2_K, IQ2, IQ1, TQ1) cut file size by roughly half versus Q4, but the quality collapse is steep and non-linear: perplexity spikes, instruction-following degrades, and hallucinations rise. A model that answers faster but wrong is not a smaller model: it is a worse one. ModelFit lists these builds for completeness but never ranks or recommends them.

Run Qwen3.5 27B Instruct on your GPU

RTX 3060
12 GB: verdict
RTX 4060
8 GB: verdict
RTX 4060 Ti
16 GB: verdict
RTX 4070
12 GB: verdict
RTX 4070 Ti SUPER
16 GB: verdict
RTX 3090
24 GB: verdict
RTX 4090
24 GB: verdict
RTX 5070 Ti
16 GB: verdict
RTX 5080
16 GB: verdict
RTX 5090
32 GB: verdict

Frequently asked questions

What is the best quantization of Qwen3.5 27B Instruct?

Q4_K_M is the default pick: 16.75 GB of weights, high: the default pick quality, fitting comfortably in 24 GB of memory (weights + 16k context KV-cache). Go Q6_K or Q8_0 if you have headroom.

How much memory does Qwen3.5 27B Instruct need?

At Q4_K_M, Qwen3.5 27B Instruct needs 16.75 GB for the weights plus ~1.0 GB of KV-cache at 16k context, about 17.8 GB total, so a 24 GB card or Mac (90% usable budget) runs it comfortably.

Should I use a Q2_K or IQ2 quant of Qwen3.5 27B Instruct?

No. Qwen3.5 27B Instruct at 2 bits per weight is a visibly worse model. Quality collapse at that bitrate is steep, not gradual. If only a 2-bit build fits your memory, run a smaller model at Q4_K_M instead. ModelFit lists these builds but never recommends them.

Cite this page

ModelFit: Qwen3.5 27B Instruct quantization comparison (real GGUF file sizes).
https://modelfit.io/quant-compare/qwen3.5-27b (data probed 2026-09-02, CC BY 4.0).