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

Can I run REDCELL-26B-A4B-OSINT-Cyber-APEX-GGUF on an RTX 5090?

Yes, just

Its Q8_0 file (26.9 GB) fits in the 28.8 GB an RTX 5090 can use, with room for ≈ 18K tokens of context.

Full analysis of REDCELL-26B-A4B-OSINT-Cyber-APEX-GGUF Everything that runs on RTX 5090 Get the Q8_0 file on Hugging Face
ollama run hf.co/terrorswift/REDCELL-26B-A4B-OSINT-Cyber-APEX-GGUF:Q8_0

Files and context on RTX 5090

Each file of REDCELL-26B-A4B-OSINT-Cyber-APEX-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 5090 Context (16-bit)Context (8-bit)
F1650.5 GBDoesn't fit——
Q8_026.9 GBTight18K41K
REDCELL-26B-A4B-OSINT-Cyber-APEX-I-Balanced20.7 GBFits92K189K
REDCELL-26B-A4B-OSINT-Cyber-APEX-Balanced20.7 GBFits92K189K
REDCELL-26B-A4B-OSINT-Cyber-APEX-I-Quality19.7 GBFits103K212K
REDCELL-26B-A4B-OSINT-Cyber-APEX-Quality19.7 GBFits103K212K
REDCELL-26B-A4B-OSINT-Cyber-APEX-I-Compact14.7 GBFits164K256K
REDCELL-26B-A4B-OSINT-Cyber-APEX-Compact14.7 GBFits164K256K
REDCELL-26B-A4B-OSINT-Cyber-APEX-Mini12.3 GBFits192K256K

REDCELL-26B-A4B-OSINT-Cyber-APEX-GGUF 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.