`lorivo deploy`

Validates, uploads, and deploys a LoRA adapter in one command.

Usage

lorivo deploy <dir> [options]

Option	Default	Description
`<dir>`	required	Directory with `adapter_config.json` and `adapter_model.safetensors`
`-n, --name <name>`	from `adapter_config.json` or dir	Adapter name (lowercase alphanumeric + dashes)
`-b, --base <id>`	`ADAPTERS_DEFAULT_BASE` or `qwen3-4b`	Base model id
`-o, --org <org>`	your org	Org name (admin keys only; org keys always use their own)

What it does

Validates the directory (see below).
Registers the adapter, or reuses the existing adapter with the same name, so the next deploy becomes v2, v3, ….
Uploads every file in the directory as a new version.
Deploys the version onto the GPU node and prints your inference endpoint.

Directory requirements

adapter_config.json with LoRA rank 1–64 and a base_model_name_or_path in the same family as --base
adapter_model.safetensors with a valid safetensors header
At most 10 files, at most 2 GiB total
No code or legacy-checkpoint formats (.bin, .pt, .pth, .ckpt, .npz, .onnx, .pkl, .pickle, .py, .so, .dylib)

Examples

lorivo deploy ./my-lora --name my-lora

admin key, targeting another org

lorivo deploy ./my-lora --name my-lora --org acme-ai

output

Validated ./my-lora - rank 8, 2 file(s), 1.7 MiB
Registered my-lora
Uploaded v1 in 2.1s (sha256 3f9e1c2a4b5d…)
Deployed my-lora v1 (LOADED)

Live at:  https://lorivo.dev/v1/responses
Model:    qwen3-4b:my-lora

Under the hood

Step	REST call
Find or register adapter	`GET /api/adapters`, `POST /api/adapters`
Upload files	`POST /api/adapters/:id/versions` (multipart)
Deploy	`POST /api/adapters/:id/deploy`

All steps are documented under Adapters & versions.