SmileyTatsu

SmileyGPT 1.3.0 Alpha 1. Guess who is back. Back again.

Added 2025-06-17 16:52:09 +0000 UTC

Okay so... Hi! I revived just to keep slaving myself for you to be able to coom! This time updating SmileyGPT and testing it on Claude 4 (opus).

```

- Added `🔞 | Enhance Physical Description`. Enabled by default.

- Added `🍂 | Colored Messages` mutator.

- Divided `🔞 | Enhance NSFW/NSFL` prompt into: `🔞 | Enhance NSFW` and `🔞 | Enhance NSFL` in order to save on tokens + be able to describe things better.

- Edited `🍂 | Message Handler` in order to be more intuitive to use and to make it easier for the AI to interpret. Default settings are recommended.

- Edited `🍂 | SFX-Enhancer` to enhance SFX in multiple ways.

- Edited `💊 | Clau's role (world)` in order to make "idle" characters make more presence when needed.

- Edited `📌| Stylizing` to make it more straight to the point.

- Edited prefill to adapt to the new tweaks.

- Removed useless note on `💊 | Assistant Instructions` that recommended disabling it.

- Changed `Reasoning Effort` option to "Auto" in Claude models.

- Small tweaks over the general JB improving token usage + explanation.

```

I edited A LOT of things, and even if most are minor things it has proven to affect the AI output in a big way. By default, the model is set on `Sonnet 3.7 Latest` as is the model I recommend for price/quality in the Claude family. `Opus 4.0` may be better but I haven't managed to notice a big difference in quality. The only thing I notice is that Sonnet 3.7 loves to go straight to the point and Opus 4.0 take it's time on things.

Remember to not share this version as it's not a final product and may not be optimal to use. This is only for testing/having "early" access to things. I am NOT giving support to people who are not in the Discord channel (or Patreon messages in this case).