Model Anatomy See what's inside open-weight AI models

Can I run Qwen3.8-27B on an RTX 4090?

Yes, with a GGUF version

The published weights are too big, but the Q4_K_M file from ggml-org/Qwen3.8-27B-GGUF (19.0 GB) fits, with room for ≈ 37K tokens of context.

Full analysis of Qwen3.8-27B Everything that runs on RTX 4090 Get the Q4_K_M file on Hugging Face
ollama run hf.co/ggml-org/Qwen3.8-27B-GGUF:Q4_K_M

Files and context on RTX 4090

Each file of Qwen3.8-27B (and of GGUF versions the site has analyzed), and how much context fits next to it, with the context cache in 16-bit or 8-bit.

FileSizeOn RTX 4090 Context (16-bit)Context (8-bit)
As published55.6 GBDoesn't fit——
BF16ggml-org/Qwen3.8-27B-GGUF53.8 GBDoesn't fit——
Q8_0ggml-org/Qwen3.8-27B-GGUF28.6 GBDoesn't fit——
Q4_K_Mggml-org/Qwen3.8-27B-GGUF19.0 GBTight37K74K
mtp-Qwen3.8-27B-BF16ggml-org/Qwen3.8-27B-GGUF5.95 GBFits231K256K
IQ4_XSbyteshape/Qwen3.8-27B-GGUF13.1 GBFits125K249K
IQ3_Sbyteshape/Qwen3.8-27B-GGUF11.0 GBFits155K256K
IQ3_XSbyteshape/Qwen3.8-27B-GGUF10.4 GBFits165K256K
IQ3_XXSbyteshape/Qwen3.8-27B-GGUF9.92 GBFits172K256K
IQ2_XXSbyteshape/Qwen3.8-27B-GGUF8.84 GBFits188K256K

Qwen3.8-27B on other hardware

Popular models that run on RTX 4090

Assumes 90% of the card's memory is usable and at least 4K tokens of context for one conversation, with the context cache in 16-bit unless noted. Fits leaves at least 20% of that memory for context; Tight leaves less.