Update README.md

README.md

@@ -9,7 +9,17 @@ tags:
 
 EXL3 quants of [gemma-4-12B-it](https://huggingface.co/google/gemma-4-12B-it)
 
-⚠️ Requires ExLlamaV3 v0.0.40 (or v0.0.39 `dev` branch)
+mul1 quants (recommended, improved performance on exllamav3>=v1.0.0):
+
+[3.00 bits per weight](https://huggingface.co/turboderp/gemma-4-12B-it-exl3/tree/3.00bpw_mul1)    
+[3.50 bits per weight](https://huggingface.co/turboderp/gemma-4-12B-it-exl3/tree/3.50bpw_mul1)    
+[4.00 bits per weight](https://huggingface.co/turboderp/gemma-4-12B-it-exl3/tree/4.00bpw_mul1)    
+[4.50 bits per weight](https://huggingface.co/turboderp/gemma-4-12B-it-exl3/tree/4.50bpw_mul1)    
+[5.00 bits per weight](https://huggingface.co/turboderp/gemma-4-12B-it-exl3/tree/5.00bpw_mul1)    
+[6.00 bits per weight](https://huggingface.co/turboderp/gemma-4-12B-it-exl3/tree/6.00bpw_mul1)    
+[8.00 bits per weight / H8](https://huggingface.co/turboderp/gemma-4-12B-it-exl3/tree/8.00bpw_mul1)
+
+MCG quants (legacy):
 
 [2.00 bits per weight](https://huggingface.co/turboderp/gemma-4-12B-it-exl3/tree/2.00bpw)    
 [2.50 bits per weight](https://huggingface.co/turboderp/gemma-4-12B-it-exl3/tree/2.50bpw)