Update README.md

README.md

@@ -2,15 +2,17 @@
 pipeline_tag: text-generation
 ---
 
-Collection of resources and models for storytelling and roleplay. Updated Mar 2025.
+Good story telling models that can fit in an RTX 3060 12GB. Updated July 2025.
 
 Some notes on best usage:
+- dont underestimate the original instruct models, esp from mistral
 - dont waste time on sampler settings; use recommended and optimize the prompt
 - don't "overparameterize" by writing too long a prompt
 - model size/intelligence is important but they are just mimics, the dataset is very important
 
 # model ranking
 
+- **Winner**: [mistralai/Mistral-Small-3.1-24B-Instruct-2503](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503)
 -  [Sao10K/MN-12B-Lyra-v4](https://huggingface.co/Sao10K/MN-12B-Lyra-v4)
 -  [MarinaraSpaghetti/NemoMix-Unleashed-12B](https://huggingface.co/MarinaraSpaghetti/NemoMix-Unleashed-12B)
 -  [Sao10K/Fimbulvetr-11B-v2](https://huggingface.co/Sao10K/Fimbulvetr-11B-v2)