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

Can I run Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-NM-DAU-NEO-MTP-GGUF on an RTX 3090?

Yes, just

Its Q5_K_S file (20.6 GB) fits in the 21.6 GB an RTX 3090 can use, with room for ≈ 12K tokens of context.

Full analysis of Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-NM-DAU-NEO-MTP-GGUF Everything that runs on RTX 3090 Get the Q5_K_S file on Hugging Face
ollama run hf.co/DavidAU/Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-NM-DAU-NEO-MTP-GGUF:Q5_K_S

Files and context on RTX 3090

Each file of Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-NM-DAU-NEO-MTP-GGUF (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 3090 Context (16-bit)Context (8-bit)
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-TOOLS-NM-DAU-NEO-MAX-MTP-Q8_030.2 GBDoesn't fit——
Q8_030.2 GBDoesn't fit——
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-Q8_029.8 GBDoesn't fit——
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-TOOLS-NM-DAU-NEO-MTP-Q8_029.0 GBDoesn't fit——
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MTP-Q8_029.0 GBDoesn't fit——
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-Q8_028.6 GBDoesn't fit——
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-TOOLS-NM-DAU-NEO-MAX-MTP-Q6_K24.0 GBDoesn't fit——
Q6_K24.0 GBDoesn't fit——
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-Q6_K23.6 GBDoesn't fit——
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-TOOLS-NM-DAU-NEO-MTP-Q6_K22.4 GBDoesn't fit——
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MTP-Q6_K22.4 GBDoesn't fit——
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-Q6_K22.1 GBDoesn't fit——
Q5_K_M21.2 GBDoesn't fit4K8K
Q5_K_S20.6 GBTight12K24K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MTP-Q5_K_M19.5 GBTight28K57K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MTP-Q5_K_S19.0 GBTight37K74K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-TOOLS-NM-DAU-NEO-MAX-MTP-Q4_K_M18.5 GBTight44K88K
Q4_K_M18.5 GBTight44K88K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-Q4_K_M18.0 GBTight51K101K
Q4_K_S17.5 GBTight58K116K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-TOOLS-NM-DAU-NEO-MAX-MTP-IQ4_XS17.0 GBFits66K131K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-MTP-IQ4_XS17.0 GBFits66K131K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-TOOLS-NM-DAU-NEO-MTP-Q4_K_M16.8 GBFits69K138K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MTP-Q4_K_M16.8 GBFits69K138K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-IQ4_XS16.6 GBFits72K145K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-Q4_K_M16.5 GBFits73K146K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MTP-Q4_K_S15.8 GBFits84K168K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-TOOLS-NM-DAU-NEO-MTP-IQ4_XS15.3 GBFits91K183K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MTP-IQ4_XS15.3 GBFits91K183K
IQ4_XS15.1 GBFits95K190K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-MTP-IQ3_M14.5 GBFits103K206K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-IQ3_M14.1 GBFits110K220K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MTP-IQ3_M12.8 GBFits129K256K
IQ3_M12.6 GBFits132K256K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-MTP-IQ2_M12.1 GBFits139K256K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MAX-IQ2_M11.7 GBFits146K256K
Qwen3.8-27B-TTURBO-Fable-C-Fusion-709-L-Uncen-NM-DAU-NEO-MTP-IQ2_M10.2 GBFits167K256K
IQ2_M10.0 GBFits170K256K

Qwen3.8-27B-TWIN-TURBO-Fable-Cold-Fusion-709-L-Uncensored-NM-DAU-NEO-MTP-GGUF on other hardware

Popular models that run on RTX 3090

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.