Update README to clarify latency benefits for self-hosted models

Author: robertpiosik

README.md

@@ -39,7 +39,7 @@ Code Web Chat (CWC) is a free and open-source, privacy-first **AI coding toolkit
 
 ## Cost efficiency
 
-First-class support for [prompt caching](https://developers.openai.com/api/docs/guides/prompt-caching/) by ordering the selected files by update recency. Make your chatbot quotas last longer or lower token costs by up to 90% when calling APIs. Forget about slow prompt processing with local inference.
+First-class support for [prompt caching](https://developers.openai.com/api/docs/guides/prompt-caching/) by ordering the selected files by update recency. Make your chatbot quotas last longer or lower token costs by up to 90% when calling APIs. Minimize latency of self-hosted models.
 
 ## Enabling autofill