>>715735206
keep in mind when using openrouter/chutes that you might accidentally land on non-official api, aka deepseek that is ran distilled which can degrade quality up to 75% depending on how low quants are

make use of token caching, if you have anything setup in sillytavern that inserts new things or changes the very beginning of chat (start prompts) then it will invalidate token cache, and cache hits on official api reduce price by 4x per token