How to download Capri?

I get asked this by hundreds of ppl daily, so let's clear up the most popular questions.

Capri is still under development and has a bunch of unfinished features that need tuning, but I'm gonna drop it completely on my GitHub.

How much does it cost?

The project is open-source and totally free. It doesn't require any third-party APIs like Claude or ElevenLabs. If you wanna support the development, you can head over to my Boosty and drop a small donation to help the project grow.

What are the system specs?

I run a local Qwen 3.5 model with 9B parameters. It eats up all the VRAM on my RTX 4070—exactly 12 gigs. If you don't have that much memory, you can always use lighter models like Qwen 4B or Qwen 2B. Some speech recognition and generation models use both VRAM and RAM, so you'll need to find a sweet spot for your hardware. Detailed specs will be on GitHub.

Can Capri speak my language?

Yeah, I'm integrating a multilingual model, so you'll be able to chat with it in almost any language. Keep in mind though, you'll get the absolute best quality with the English voiceover.

When is the release date?

The big question! I'm planning to finish the first alpha version this week, maybe even in the next two days. Make sure to follow and stay tuned for updates.

All links are in my bio!