Browse & discover
The in-app model browser connects to Hugging Face and filters for models compatible with Nimbus8's runtimes — GGUF and MLX formats. You can search by name, filter by size, and see which quantization variants are available. Every listing shows the model's parameter count, file size, and which modules it works with.
Install & manage
Tap to download. Downloads are resumable — if you lose connection or close the app, it picks up where it left off. Once installed, models appear in the model picker across all compatible modules. Swap between them with a single tap. Uninstall reclaims the disk space immediately.
The model registry tracks every installed model's repo, filename, SHA256 hash, and last-loaded timestamp. TOFU (trust on first use) verification ensures the file hasn't been tampered with after initial download.
Gated models
Some models on Hugging Face (Llama, Gemma, etc.) require accepting a license before download. Paste your Hugging Face token in Settings to access gated repos. The token is stored in the iOS Keychain and never logged or transmitted anywhere except to Hugging Face's download API.
Bring your own
If you have a GGUF file from another source, you can import it directly via the Files app or AirDrop. Nimbus8 will detect the format, run a benchmark, and add it to your local registry. No Hugging Face account required.