We're back!

gemma 4 31b

You can use Gemma 4 Thinking template if you want thinking!

(Disclaimer: This was not trained with the empty think tags. You might want to test w/ and w/o the <|turn>model\n<|channel>thought\n<channel|> prefill for no-think RP.)

Downloads last month: 5,975

GGUF

Model size

31B params

Architecture

gemma4

Hardware compatibility

2-bit

3-bit

4-bit

5-bit

6-bit

8-bit

Inference Providers NEW

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