Start with the built-in sample or your own folder.
On an empty Overview, select Try sample model family. Tensor Archive generates a small synthetic base model and three related adapter deployments, runs them through the real local archive pipeline, verifies the result and opens the measured family. It needs no account, model download or internet connection.
To prove the same path on real public weights, choose Google BERT Tiny. The preset is an official Apache 2.0 Google checkpoint fixed to one immutable Hugging Face commit. Analyze reads its metadata first; Archive then downloads the complete 53,265,006-byte snapshot. The synthetic family demonstrates cross-version savings; the Google preset demonstrates real-model compatibility.
-
1
Try the built-in family
Choose Try sample model family to see real Versions, Files, Tensors, Activity and Storage results immediately.
-
2
Or select Add model family
From Overview, choose Local folder, then use Choose folder.
-
3
Analyze before writing
Select Analyze source. Review the detected type, files, source bytes and storage plan.
-
4
Name and archive
Give the family a clear catalog name and select Archive folder. Progress appears as a local job.
-
5
Verify, then test a restore
Open the family, run Verify, restore a version, and compare or open it before deleting any source copy.
Archiving creates a managed copy; it does not move or delete your source. Keep the original until Verify succeeds and you have tested an exact restore.
Understand what each storage number means.
Tensor Archive separates the size your versions represent from the bytes physically required to keep them.
The total size if each archived version were kept as its own self-contained copy.
The bytes currently occupied by the family inside Tensor Archive's local storage.
The repeated model data that no longer needs another physical copy inside the archive.
The additional physical storage required after content already present in the family is reused.
71.2% is a measured result from one related model family, not a promise for every folder. Families that share substantial content can save a lot; unrelated or already compressed files may save little.
Archive the base first. Add related versions after it.
The family relationship matters: it lets the product show every version while measuring only the new physical bytes each one adds.
Checkpoints and model directories
Choose a folder using the native macOS picker. Tensor Archive receives temporary access to that selection, checks it locally and rejects unsafe links or special files.
Attach a LoRA to its archived base
Archive the base checkpoint first. When an adapter is detected, choose the correct base. If its declared base does not match, the app requires an explicit confirmation before continuing.
Resolve a public repository revision
Enter owner/model and a branch, tag or commit. Analyze reads the public file inventory and may inspect a small adapter configuration; it does not download or execute model weights. Archive then resolves an immutable commit and stores the complete public snapshot locally. The Google BERT Tiny preset offers a small official first source.
Use one family for genuinely related models
Group a base, its adapters and its real revisions together. Separate unrelated architectures or projects so that the catalog stays meaningful.
Inspect, prove, recover or carry it elsewhere.
Reads the archived payload and creates a durable receipt. It verifies exact bytes and relationships—not model quality, accuracy or inference behavior.
Creates a byte-for-byte output under ~/Library/Application Support/ai.tensorarchive.local/outputs/restores/. The current beta does not yet offer a custom destination picker.
.tensorarchive fileSelect Download when the job completes. Version 0.1.3 saves it in Downloads, avoids overwriting an existing file and reveals it in Finder. Importing that file through the desktop interface is not available in this beta yet.
Browse file paths, sizes and checksums. Safetensors entries can also be inspected and individually extracted without restoring the whole version.
Archive, verify, restore, export and extraction run as local jobs with progress, status and completion receipts.
Delete originals only after you prove recovery.
Tensor Archive deliberately does not remove source data. That final storage decision stays with you.
1Archive the folder and wait for successful completion.
2Verify the archived version and retain its receipt.
3Restore it to a separate output and confirm your workflow can use it.
4Back it up externally if loss of this Mac would matter.
5Delete the old source manually only when you are satisfied with those checks.
Local deduplication saves disk space, but it does not protect against disk failure, theft or accidental removal of the application data directory. Keep another copy of irreplaceable model data.
Where Tensor Archive is most useful today.
- A base model with several LoRAs or adapters.
- Related checkpoints and training revisions.
- Understanding the true incremental cost of each version.
- Freeing local SSD space after verified recovery tests.
- Auditing files, hashes, relationships and tensor inventories.
- Exact offline restores from a local archive.
- A training or inference accelerator.
- Quantization, pruning or a model conversion tool.
- A way to improve model quality or accuracy.
- A guaranteed 71% saving for every collection.
- A cloud backup or multi-device sync service in this beta.
- A replacement for an independent backup of important data.
Tensor Archive Local 0.1.3 supports Apple silicon Macs running macOS 13 or later. The built-in synthetic family is included. Public Hugging Face repositories and local folders are also available. Private repository authentication, S3, cloud sync, team catalogs, Windows and Linux are not available in this build yet.
Your local models stay local.
No account is required
Local folders and archive contents are not uploaded to Tensor Archive. Network access is used when you explicitly choose Hugging Face and when the app checks tensorarchive.ai for product updates.
One private application directory
The catalog, stored content, receipts and generated outputs live under
~/Library/Application Support/ai.tensorarchive.local/.
Removing that directory removes the local archive too.
Pro creation during the preview
This evaluation build can analyze and create archives until September 1, 2026. After the preview, Viewer / Restore remains available for existing archives; creating or extending families requires Local Pro.
Automatic check, manual replacement
Version 0.1.3 checks the official update manifest. While the macOS preview remains unsigned, download and install the newer DMG manually when the app reports an update.