Update README.md

README.md

@@ -151,6 +151,20 @@ Record so far (mine): 12k output (coherent) with 4k context limit.
 For some AI apps use of the Jinja Template (embedded in the GGUFs) may not work, and you need to manual select/use "ChatML" template
 in your AI/LLM app.
 
+<B>Quant Choice Notes:</b>
+
+This model shows much stronger detail, generation and thoughts/reasoning as you go up in quant(s).
+
+In terms of "reasoning/thinking" length this can be HALVED for some "problems" even if you go from Q2k up to Q3KM.
+
+I.E: It figures out the solution to the problem faster.
+
+Likewise, detail in output as well as detail in reasoning will be deeper and stronger.
+
+With that in mind, even Q2k (the smallest/lowest regular quant) is potent.
+
+Also, the same quant in "Imatrix" maybe even stronger than the regular version.
+
 <b>Optional : Rocket Fuel for Thought</b>
 
 This is an optional system prompt you can use to enhance both "thinking/reasoning" and "output".