Rent a GPU

Qwen2.5 Coder 32B hardware requirements

32.8B parameters. Needs about 22.6 GB at Q4_K_M with 8K context.

Parameters33BDense
Q4_K_M · 8K22.6 GBthe usual download
FP16 · 8K68.3 GBfull precision
Runs well on16/51GPUs and Macs we track

Memory needed at each quantization

Weights, KV cache and runtime overhead, at 8K and 32K tokens of context.

QuantQuality8K context32K context
FP16Full quality68.3 GB74.8 GB
Q8_0Near-lossless37.6 GB44 GB
Q6_KExcellent29.6 GB36.1 GB
Q5_K_MVery good26.1 GB32.5 GB
Q4_K_MGood, most popular22.6 GB29.1 GB
Q3_K_MNoticeably degraded18.8 GB25.2 GB
Q2_KHeavily degraded15 GB21.5 GB

Run it with Ollama

Downloads Q4_K_M by default.

ollama run qwen2.5-coder:32b

Model card

Qwen/Qwen2.5-Coder-32B-Instruct on Hugging Face: licence, prompt format and GGUF builds.

Run it in the cloud

Run Qwen2.5 Coder 32B on a rented GPU today.

It needs about 22.6 GB at Q4_K_M: a rented RTX 3090 / RTX 4090 runs it for about $0.18 to $0.89 an hour.

RunPodRent a GPU on RunPodVast.aiRent a GPU on Vast.ai

Every GPU and Mac checked

Sorted by memory. Bars show how much of each card Qwen2.5 Coder 32B fills at the best quant that fits.

16 Runs well

NameMemory usedBest quantSpeed

17 Fits with heavy quantization

NameMemory usedBest quantSpeed

18 Too big

NameMemory usedBest quantSpeed

Other Qwen models