Download Ollaya
One binary for Linux and macOS, or a Docker image.
Coming soon — first release in progress. Nothing is downloadable yet. Star / watch on GitHub to hear about the first release.
Install with one command
Coming sooncurl -fsSL https://ollaya.cobanov.dev/install.sh | shUntil the first release ships, the script only prints a notice and exits.
Planned requirements
- Runs on the CPU out of the box (ONNX Runtime).
- NVIDIA GPUs are accelerated with CUDA.
- Roughly 0.65–1.7 GB of disk per model, depending on model and precision.
Final requirements will be published with the first release.
Install with one command
Coming sooncurl -fsSL https://ollaya.cobanov.dev/install.sh | shUntil the first release ships, the script only prints a notice and exits.
Planned requirements
- Accelerated with Core ML; runs on the CPU as a fallback.
- Roughly 0.65–1.7 GB of disk per model, depending on model and precision.
Final requirements will be published with the first release.
CPU only
Coming soondocker run -d -p 11435:11435 -v ollaya:/root/.ollaya --name ollaya ghcr.io/cobanov/ollayaNVIDIA GPU
Coming soondocker run -d --gpus=all -p 11435:11435 -v ollaya:/root/.ollaya --name ollaya ghcr.io/cobanov/ollayaRequires the NVIDIA Container Toolkit on the host.
Run a model
Coming soondocker exec -it ollaya ollaya run laya --preset triage "I was charged twice this month."The image will be published to ghcr.io/cobanov/ollaya with the first release.