💦 CREATE AND FUCK YOUR OWN AI CUMSLUT 🔥 TRY NOW ♥
Create AI Cumsluts 💦
riko_project_v2_patreon.7z 163.8 MB
September 4, 2026

Riko Project V2 | 2026-09-04

Riko Project V2 | Out 2026-09-04

Hey guys! Big one. Version 2 of the Riko Project is here.

I rebuilt the whole thing from the ground up. The old version was getting harder and harder to add to, especially when it came to anything live. V2 fixes that, so adding new features should be a lot faster from here on out.

What's new

  • She moves better now. The new architecture lets her pick her own expressions and animations while she talks. She writes them directly into her reply, and the model plays them. Smug faces, waving, all of it.

  • New UI. It's actually wired now.

  • Custom backgrounds. Put her in whatever room you want. There's also a lighting sandbox where you can drag the lights around until everything looks right, then copy the result into your setup.

  • New voice option: Fish Audio. You can now use Fish Audio online instead of running GPT-SoVITS locally, so you don't need a GPU if you don't have one. GPT-SoVITS is still available if you'd rather clone the voice on your own machine. Both options work.

    👉 Sign up for Fish Audio here: https://fish.audio/?fpr=rayen10

  • Way easier to set up. Run setup.bat, paste in your keys, and you're done. There's also a check_setup.bat that tests every part of the system and tells you exactly what's broken and how to fix it.

  • More customizable. Her personality, name, voice, model, room, lighting, and more are all controlled through plain config files with comments explaining what everything does. You can also swap in any VRM model you like.

  • Desktop mode: A transparent window with her floating over whatever you're doing.

Still in there, and better

  • VR: Open it on a Quest and she'll be standing right in front of you, at scale.

  • Tool calling: She can actually do things, not just talk about them.


The download is up now. The full setup guide is included in the files. Start with SETUP_GUIDE.md, and if anything acts up, run check_setup.bat.

Let me know what you think, and what you'd like to see next!

Rayen

Comments (9)

B
Bizzyjax · September 10, 2026

hey thank you for the update.

o
omar falioun · September 8, 2026

can you add cloude api support please ?

R
Rekcah Keeg · September 7, 2026

I have a 3090 is there a way I can run a local llm to avoid open ai API fees? I am noticing the API fees add up faster than expected. I have also tried to take a paid for model and convert the FBX files to vrm in blender has not been great so far.

D
Doktor Sebo · September 5, 2026

Desperate times call for JustRayen. Go watch him on Twitch btw

T
Troy Flater · September 4, 2026

good job -- i gave the character_config.yaml a little tweak

d
dazedpilot · September 4, 2026

this install was so much easer and works well

S
Sim73 · September 4, 2026

Hi, I strongly recommend using F5-TTS instead of those TTS systems. The quality is great, vram occupation is around 1 Gbyte, and execution time is very good. It is also very easy to use with Python. The second best is ElevenLabs: quality is the best but the cost can be significant; that said, the Turbo model is good and relatively cheap.

T
Tim Walter · September 5, 2026

I use IndexTTS2.5, because SoVITS dont work on my 5070ti.

M
Megan Golder · September 4, 2026

goated