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

Can I run Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-GGUF-NInfer on an RTX 4090?

Yes, just

Its Qwen3.8-27B-Coder390-EfficientThink-Q4LynnStyle-Q8MTP file (19.8 GB) fits in the 21.6 GB an RTX 4090 can use, with room for ≈ 24K tokens of context.

Full analysis of Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-GGUF-NInfer Everything that runs on RTX 4090 Get the Qwen3.8-27B-Coder390-EfficientThink-Q4LynnStyle-Q8MTP file on Hugging Face
ollama run hf.co/nerkyor/Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-GGUF-NInfer

Files and context on RTX 4090

Each file of Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-GGUF-NInfer (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)
Q8_029.1 GBDoesn't fit——
Q6_K23.2 GBDoesn't fit——
Qwen3.8-27B-Coder390-EfficientThink-Q4LynnStyle-Q8MTP19.8 GBTight24K48K
Qwen3.8-27B-Coder390-EfficientThink-Q4LynnStyle-MTP19.6 GBTight27K54K
Qwen3.8-27B-Coder390-EfficientThink-Q3LynnStyle-Q8MTP17.5 GBTight59K117K
Qwen3.8-27B-Coder390-EfficientThink-Q3LynnStyle-MTP17.3 GBTight62K123K
Qwen3.8-27B-Coder390-EfficientThink-Q2LynnStyle-Q8MTP13.5 GBFits119K237K
Qwen3.8-27B-Coder390-Q3LynnStyle16G-MTP13.3 GBFits122K243K
Qwen3.8-27B-Coder390-EfficientThink-Q2LynnStyle-MTP13.3 GBFits122K243K

Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-GGUF-NInfer 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.