Repetere AI 1.0 Sovereign Workstation
|
Native C++20 / GLFW / OpenGL 3 Host
● System Requirements · consumer hardware
If it runs a game, it runs Repetere.
The host sips under 90 MB. Your GPU decides model size and speed — pick a tier below and the host tunes each studio to it.
Minimum · text studios
Chat, Code, Story
- OS: Windows 10/11 64-bit · Ubuntu 22.04+
- CPU: 4 cores / 8 threads, AVX2
- RAM: 8 GB (host uses <90 MB)
- GPU: Any Vulkan 1.2 iGPU — Intel UHD / Vega
- Disk: 20 GB + 4–8 GB per text model
Recommended · all studios
Full five-studio rig
- CPU: 6+ cores, modern Ryzen / Core i5+
- RAM: 16–32 GB DDR4/DDR5
- GPU: RTX 3060 12 GB / RX 6700 XT / Arc A770
- Engine: CUDA on RTX · Vulkan on Radeon/Arc
- Disk: 100 GB NVMe vault (models + renders)
Enthusiast · diffusion + music
Batch rendering
- GPU: RTX 4070+ / 16 GB VRAM preferred
- RAM: 32 GB+ for large batches
- Disk: 250 GB+ vault, SDXL + stems
- Extra: Second drive for exports
- Note: CPU-only diffusion works, slowly
Engine matrix: CUDA & Vulkan hot-swap
NVIDIA RTX (Turing+)CUDA · fastest
AMD Radeon (RX 5000+)Vulkan · full
Intel Arc / integratedVulkan · efficient
No discrete GPUCPU fallback · text OK
Switch per studio without restarting. First boot auto-benchmarks and recommends the fastest binding.