Downloads for PC
Want to run models locally on your own GPU? Install these, in order.
1 · Frame for Windows
The desktop app (frame.exe) with built-in web UI. Installs to C:\Frame\.
Frame for Linux (cloud GPU)
Run Frame on a rented GPU server (RunPod / Vast.ai) and create from the web studio on any device.
⬇ Download Linux build 1.7.63 · Image models
SDXL base 1.0 → checkpoints/, or Flux.1 schnell/dev FP8 → diffusion_models/ + VAE + text encoders. Optional LoRAs → loras/.
4 · Video models
Wan 2.2 TI2V 5B for 12GB cards; I2V/T2V 14B FP8 for 16GB+. Plus Wan VAE and umt5 text encoder.
Wan 2.2 for ComfyUI5 · Ollama (prompt assistant)
Install Ollama, then e.g. ollama pull qwen3.5:9b (smaller model for 8GB).
6 · Whisper (optional)
Voice prompts. Added automatically under C:\Frame\whisper on first use.
Android app
Phone companion — connects to your PC or your cloud gallery. Test build (debug-signed).
Download APKNo suitable GPU? Skip steps 2–5 and connect Frame to a remote ComfyUI (cloud GPU) in Settings → ComfyUI URL, or just use the web app.