Can I run gemma-4-26B-A4B-it on an RTX 6000 Ada?

Yes, with a GGUF version

The published weights are too big, but the UD-Q8_K_XL file from unsloth/gemma-4-26B-A4B-it-GGUF (27.6 GB) fits, with room for ≈ 181K tokens of context.

Full analysis of gemma-4-26B-A4B-it Everything that runs on RTX 6000 Ada Get the UD-Q8_K_XL file on Hugging Face

ollama run hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:UD-Q8_K_XL

Files and context on RTX 6000 Ada

Each file of gemma-4-26B-A4B-it (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.

File	Size	On RTX 6000 Ada	Context (16-bit)	Context (8-bit)
As published	51.6 GB	Doesn't fit	—	—
BF16unsloth/gemma-4-26B-A4B-it-GGUF	50.5 GB	Doesn't fit	—	—
UD-Q8_K_XLunsloth/gemma-4-26B-A4B-it-GGUF	27.6 GB	Fits	181K	256K
Q8_0unsloth/gemma-4-26B-A4B-it-GGUF	26.9 GB	Fits	190K	256K
UD-Q6_K_XLunsloth/gemma-4-26B-A4B-it-GGUF	23.3 GB	Fits	232K	256K
UD-Q6_Kunsloth/gemma-4-26B-A4B-it-GGUF	23.2 GB	Fits	234K	256K
UD-Q5_K_XLunsloth/gemma-4-26B-A4B-it-GGUF	21.2 GB	Fits	256K	256K
UD-Q5_K_Munsloth/gemma-4-26B-A4B-it-GGUF	21.2 GB	Fits	256K	256K
UD-Q5_K_Sunsloth/gemma-4-26B-A4B-it-GGUF	18.9 GB	Fits	256K	256K
UD-Q4_K_XLunsloth/gemma-4-26B-A4B-it-GGUF	17.0 GB	Fits	256K	256K
UD-Q4_K_Munsloth/gemma-4-26B-A4B-it-GGUF	16.9 GB	Fits	256K	256K
MXFP4_MOEunsloth/gemma-4-26B-A4B-it-GGUF	16.6 GB	Fits	256K	256K
UD-Q4_K_Sunsloth/gemma-4-26B-A4B-it-GGUF	16.5 GB	Fits	256K	256K
UD-IQ4_NLunsloth/gemma-4-26B-A4B-it-GGUF	13.6 GB	Fits	256K	256K
UD-IQ4_XSunsloth/gemma-4-26B-A4B-it-GGUF	13.6 GB	Fits	256K	256K
UD-Q3_K_XLunsloth/gemma-4-26B-A4B-it-GGUF	12.9 GB	Fits	256K	256K
UD-Q3_K_Munsloth/gemma-4-26B-A4B-it-GGUF	12.7 GB	Fits	256K	256K
UD-IQ3_XXSunsloth/gemma-4-26B-A4B-it-GGUF	11.4 GB	Fits	256K	256K
UD-IQ3_Sunsloth/gemma-4-26B-A4B-it-GGUF	11.3 GB	Fits	256K	256K
UD-Q2_K_XLunsloth/gemma-4-26B-A4B-it-GGUF	10.5 GB	Fits	256K	256K
UD-IQ2_Munsloth/gemma-4-26B-A4B-it-GGUF	10.0 GB	Fits	256K	256K
UD-IQ2_XXSunsloth/gemma-4-26B-A4B-it-GGUF	9.92 GB	Fits	256K	256K

gemma-4-26B-A4B-it on other hardware

Popular models that run on RTX 6000 Ada

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.