Manual Fix a problem
A model won't download
Downloads that fail, stall, or install incorrectly — and what each error message actually means.
You picked a local model and the download failed, stopped part way, or finished and then wouldn't install.
The error you got narrows it down considerably, so start there.
What each message means
| Message | What it means | What to do |
|---|---|---|
| Certificate error — your network may use SSL inspection | A proxy is intercepting the download and its root isn't trusted | Download from a different network, or have IT trust the proxy root |
| Server error | The model host returned an error | Wait and retry — this is usually temporary and not on your side |
| Model downloaded but installation failed | The file arrived but couldn't be unpacked | Click download again. It reuses what you already downloaded rather than fetching it a second time |
| Download failed | A generic network failure | Check the connection and retry; see below if it keeps happening |
A model that says it's downloaded but won't run
If a model shows as downloaded in the picker but transcription fails with a "not downloaded" message, an interrupted download almost certainly left an empty or partial file behind, and the picker is trusting the folder rather than the contents.
Close Neato Echo
From the tray icon, so nothing is holding the file open.
Delete the model's folder
Open
%USERPROFILE%\.cache\neato-echoin File Explorer and delete the folder for that model only. Leave the others alone.Download it again
Reopen the app, go to Settings → Speech-to-Text (or Language Models) under AI Models, and download the model fresh.
If it keeps failing
Check you have the disk space
Models are large, and a download that dies part way through is usually a full disk. The default Parakeet model is about 600 MB, the recommended Qwen 3.5 4B language model about 2.5 GB, and the larger Whisper models need several gigabytes free.
Try a smaller model
If a small model downloads and a large one doesn't, that's disk space or a connection dropping over a long transfer — not a broken installation. Whisper Base, or a smaller Qwen size, are quick tests.
Try a different network
Corporate networks, VPNs and DNS filters block the model hosts more often than anything else, because the files come from
huggingface.coand GitHub rather than from us.Delete the model and start again
In the model picker, delete the partial model, then download it fresh. That clears a corrupted partial file, which retrying alone won't always fix. If the picker won't let you delete it, remove the folder by hand as described above.
On a managed network
Model downloads are the one thing the default local setup needs the internet for, so they can fail on a network where everything else works. Your IT team needs to allow:
huggingface.cocdn-lfs.huggingface.co,cdn-lfs-us-1.huggingface.cogithub.com,objects.githubusercontent.com
See For your IT team.
FAQ
Can I use Neato Echo while a model downloads?
Yes. Anything that already has its model can keep working; the mode waiting for the new model can't transcribe until it lands.
Do I need to download a model at all?
For transcription, yes — Neato Echo runs speech-to-text on your PC, and the default Parakeet model is downloaded on first run. Summaries, cleanup and the agent need a language model from Settings → Language Models. The only way around a download is to add a key under Cloud Providers and use that provider instead.
Where do models go, and how much space do they use?
In %USERPROFILE%\.cache\neato-echo. The model picker shows each one's size
before you download it, and you can delete any of them from the same place.
It downloads every time I restart.
That points at the install step failing rather than the download — the file never lands where the app expects it. Send us a debug log; see how.
Related
Still need help?
Write to [email protected] with your Windows version, your Neato Echo version (Settings → System), and what you were doing. A person reads every message.