Importable models

⚠ Check the license. Every model carries its author's license — NC models are non-commercial. Imported models are not curated or benchmarked by the plugin team; use them at your own discretion. How to import - easiest way (plugin v1.8.3.6+): Jellyfin Dashboard → Plugins → AI Upscaler → card "Import Community Model" - pick the model and the plugin downloads it, verifies the sha256 and registers it automatically. Also: the AI service dashboard (:5000 → Models → Import from OpenModelDB) imports directly, and with the optional docker7-converter image the ~600 convertible .pth models below install too (auto-converted to ONNX, sha256-pinned; Google-Drive/Mega-hosted ones via the config page's file-install form). Manual alternative: download the ONNX file below, then upload it to your AI service — it is validated and appears in the plugin's model list immediately:
curl -X POST http://<service>:5000/models/upload -F file=@model.onnx -F model_name=my-model-x4 -F scale=4
(add -H "X-Api-Token: …" if your service uses a token; requires ENABLE_MODEL_UPLOAD=true, which is the default). Alternative: place the file in the service's model directory per MODEL-HOSTING.md.