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

Can I run Arete-27B on an RTX 5090?

Yes, with a GGUF version

The published weights are too big, but the Q6_K_L file from bartowski/NousResearch_Arete-27B-GGUF (24.7 GB) fits, with room for ≈ 59K tokens of context.

Full analysis of Arete-27B Everything that runs on RTX 5090 Get the Q6_K_L file on Hugging Face
ollama run hf.co/bartowski/NousResearch_Arete-27B-GGUF:Q6_K_L

Files and context on RTX 5090

Each file of Arete-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 5090 Context (16-bit)Context (8-bit)
As published54.7 GBDoesn't fit——
BF16bartowski/NousResearch_Arete-27B-GGUF53.8 GBDoesn't fit——
Q8_0bartowski/NousResearch_Arete-27B-GGUF28.7 GBDoesn't fit——
Q6_K_Lbartowski/NousResearch_Arete-27B-GGUF24.7 GBTight59K117K
Q6_Kbartowski/NousResearch_Arete-27B-GGUF23.6 GBTight75K150K
Q6_K_Sbartowski/NousResearch_Arete-27B-GGUF22.6 GBFits90K180K
Q5_K_Mbartowski/NousResearch_Arete-27B-GGUF20.7 GBFits119K237K
Q5_K_Sbartowski/NousResearch_Arete-27B-GGUF19.3 GBFits139K256K
Q4_K_Lbartowski/NousResearch_Arete-27B-GGUF18.6 GBFits150K256K
Q4_1bartowski/NousResearch_Arete-27B-GGUF17.6 GBFits165K256K
IQ4_NLbartowski/NousResearch_Arete-27B-GGUF17.2 GBFits171K256K
Q4_K_Mbartowski/NousResearch_Arete-27B-GGUF17.2 GBFits171K256K
Q4_K_Sbartowski/NousResearch_Arete-27B-GGUF16.1 GBFits187K256K
Q4_0bartowski/NousResearch_Arete-27B-GGUF16.1 GBFits187K256K
IQ4_XSbartowski/NousResearch_Arete-27B-GGUF15.2 GBFits200K256K
IQ3_Mbartowski/NousResearch_Arete-27B-GGUF14.6 GBFits209K256K
Q3_K_Lbartowski/NousResearch_Arete-27B-GGUF13.9 GBFits220K256K
Q3_K_Mbartowski/NousResearch_Arete-27B-GGUF13.2 GBFits231K256K
IQ3_XSbartowski/NousResearch_Arete-27B-GGUF12.6 GBFits240K256K
Q3_K_Sbartowski/NousResearch_Arete-27B-GGUF12.5 GBFits241K256K
IQ3_XXSbartowski/NousResearch_Arete-27B-GGUF12.1 GBFits247K256K
Q2_Kbartowski/NousResearch_Arete-27B-GGUF10.6 GBFits256K256K
IQ2_Mbartowski/NousResearch_Arete-27B-GGUF10.3 GBFits256K256K
IQ2_Sbartowski/NousResearch_Arete-27B-GGUF9.45 GBFits256K256K
IQ2_XSbartowski/NousResearch_Arete-27B-GGUF8.85 GBFits256K256K
IQ2_XXSbartowski/NousResearch_Arete-27B-GGUF8.64 GBFits256K256K

Arete-27B on other hardware

Popular models that run on RTX 5090

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.