google / gemma-4-26B-A4B-it

exl3 8.00bpw

QUANTIZED BY: UnstableLlama

Information

⚠️ Requires ExLlamaV3 v0.0.29 (or v0.0.28 `dev` branch)
exl3 quantization of gemma-4-26B-A4B-it via exllamav3.
repo generated automatically with ezexl3.

stay tuned for measurements and possible chat template / tokenizer updates as we learn more...

Repo Data

excat model signature

CLI Download

hf download UnstableLlama/gemma-4-26B-A4B-it-exl3-8.00bpw --local-dir ./gemma-4-26B-A4B-it-exl3-8.00bpw

Downloads last month: 27

Safetensors

Model size

15B params

Tensor type

BF16

F16

I16

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for UnstableLlama/gemma-4-26B-A4B-it-exl3-8.00bpw

Base model

google/gemma-4-31B-it

Quantized

(165)

this model