FAQ.md

ODS FAQ

Release channel: the install commands on this page fetch development main, which is not signed. A signed-source path is staged in Verified Install Preview; it is not active until the first eligible immutable release is published, and historical v3.0.0 is not eligible.

Quick answers to common questions.

Looking for install/runtime troubleshooting? See TROUBLESHOOTING.md and INSTALL-TROUBLESHOOTING.md.

Hardware

What hardware do I need?

ODS picks a model for the memory it detects; the hardware table in the README shows the current picks. Roughly:

CPU only: 8-12 GB of RAM runs Qwen3.5 2B, 16-20 GB runs Qwen3.5 4B, 24 GB or more runs Qwen3.5 9B.
NVIDIA 8-16 GB of VRAM runs Qwen3.5 9B (64K context); 24-32 GB runs Qwen3.5 27B (64K).
NVIDIA 40 GB or more, AMD Strix Halo with 64-128 GB, or Apple Silicon with 48 GB or more runs Qwen3.6 35B-A3B (128K); NVIDIA 90 GB or more on x86_64 runs Qwen3 Coder Next (128K).

The installer needs 30 GB (most machines) to 150 GB (40 GB GPUs) of free disk plus room for the model, and NVIDIA needs driver 570 or newer. See Hardware Sizing for the per-tier disk and RAM checks.

How fast is it, and how many users can it serve?

ODS publishes no measured price, throughput or capacity figures. A default install serves one request at a time; see Multi-User Setup and run ods benchmark on your own hardware.

Capabilities

What can ODS do?

Out of the box:

💬 Dashboard/Portal agent chat on qualified Linux hosts and Apple Silicon Macs; Open WebUI remains the chat fallback on other hosts
🔗 API integration (OpenAI-compatible endpoints)

Whisper, Kokoro, RAG, n8n, and Open WebUI on qualified Linux hosts can be added when needed. Existing installations retain their selected services. Use --with-webui during a Linux install or add it later from the Extensions Library.

With voice profile:

🎙️ Voice conversations (speak in, speak out) with ODS Talk

With optional components:

🔒 Privacy Shield (PII redaction proxy)
🖼️ Image generation (SDXL Lightning via ComfyUI)
🔍 Local web search (SearXNG)

Is it as good as a frontier cloud model?

Local models are smaller than the largest cloud models, so expect weaker results on the hardest reasoning tasks. In exchange, inference stays on your hardware, there are no per-token fees, and you choose the model. You can route specific requests to a cloud provider when you want one (ods mode hybrid).

Privacy & Security

Is it really private?

Inference and chat history stay on your machine, and ODS collects no telemetry. Unless you configure a cloud or remote provider, prompts are not sent to model providers. By default ODS reaches the internet only to download models and images, to check GitHub for ODS releases, and to run web searches the Portal agent makes for you through its search provider. See Is my data private? for how to turn each off.

No data sent to cloud model providers unless you configure one
No usage telemetry from ODS or its bundled services
No training data contribution
Running locally can support GDPR/HIPAA programs; compliance depends on how you deploy and operate it

Can I use it with sensitive data?

Yes. Common use cases:

Legal document review
Medical record analysis
Financial data processing
Internal company communications
Client confidential work

Optional: Add Privacy Shield for automatic PII redaction as an extra layer.

What about model security?

The model server runs in a Docker container on Linux; on macOS llama-server runs natively, and on Windows AMD hosts llama-server.exe runs natively on Windows.
Inference needs no outbound network after the initial download.
You control which models to run.
The server can be air-gapped if needed.

Setup & Support

How hard is it to set up?

One command on a supported system; the installer detects your hardware, downloads a model and starts the services.

Linux/macOS:

curl -fsSL https://install.osmantic.com/ods.sh | bash

The hosted endpoint proxies the current bootstrap from repository main. Reviewed merges reach it automatically after edge-cache refresh. ODS_REF selects a compatible repository checkout. See Installer Trust to inspect the script or install an audited commit manually.

Windows:

& {
    $ErrorActionPreference = 'Stop'
    $ProgressPreference = 'SilentlyContinue'
    $odsSrc = Join-Path $env:TEMP ('ods-install-' + [guid]::NewGuid().ToString('N'))
    $odsZip = Join-Path $odsSrc 'ods-main.zip'
    New-Item -ItemType Directory -Path $odsSrc | Out-Null
    Invoke-WebRequest -UseBasicParsing 'https://github.com/Osmantic/ODS/archive/refs/heads/main.zip' -OutFile $odsZip
    Expand-Archive -LiteralPath $odsZip -DestinationPath $odsSrc
    $odsEntry = Join-Path $odsSrc 'ODS-main\install.ps1'
    if (-not (Test-Path -LiteralPath $odsEntry -PathType Leaf)) { throw 'The downloaded archive does not contain the ODS installer.' }
    Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass
    & $odsEntry
}

Do not run the curl ... | bash installer from Windows PowerShell. The Windows entry point guides Ubuntu/WSL2 preparation and requires Pixel with Hermes disabled. It installs missing WSL, Docker Desktop and Ubuntu after asking, continues by itself after the one restart, and opens Portal when done; see Windows Quickstart.

The wizard:

Detects your hardware
Recommends configuration
Downloads models
Starts services
Runs health checks

How do I get updates?

ods update refreshes the container images pinned by your installed version and recreates the containers:

ods update            # apply
ods update --dry-run  # preview

It first tries to take a snapshot (a failure is reported, not fatal) and afterwards checks that the services are running; ods rollback restores the snapshot. It does not change ODS code. Code changes, including security fixes, arrive only by re-running a newer installer; see Updating an existing installation.

How do I back up and restore my data?

Create a backup (saves user data and config to .backups/):

ods backup

Create a compressed backup:

ods backup -c

List existing backups:

ods backup -l

Verify a backup's integrity:

ods backup verify <backup_id>

Restore from a backup (interactive — lets you choose from available backups):

ods restore

Restore a specific backup by ID:

ods restore <backup_id>

Rollback after a failed update (restores the pre-update snapshot):

ods rollback

ods update normally creates a pre-update snapshot first, so ods rollback is usually available right after an update. On Portal (Pixel) installs, ordinary backups refuse to run because they can't capture Portal's state; ods backup -t config still works.

What are service templates?

Templates are curated presets that enable a group of extensions suited to a specific use case — for example, a creative-studio setup (image generation + voice) or a research workflow (RAG + web search + agents).

List available templates:

ods template list

Preview what a template will change before applying:

ods template preview <template-id>

Apply a template (enables the template's services):

ods template apply <template-id>

Applying a template only enables services — it doesn't disable anything you've already set up.

Can ODS reuse a model already running in Ollama or LM Studio?

The Linux installer can reuse a host-managed text/chat model instead of downloading and starting a duplicate GGUF with ODS's llama-server.

Interactive installs discover matching local services and ask before adopting one. Non-interactive installs require an explicit decision:

./install.sh --reuse-external-llm --non-interactive<
4328
/pre>

Or configure the endpoint and exact provider model directly:

./install.sh \
  --external-llm-url http://127.0.0.1:11434 \
  --external-llm-provider ollama \
  --external-llm-model qwen3.5:9b

The installer verifies the model and a real completion before changing the Compose topology. Open WebUI, ODS Talk, Hermes, Perplexica, Privacy Shield, and Token Spy then use the container-safe external endpoint. ODS does not stop the external process and does not activate local catalog models while that backend is selected.

To return an existing installation to ODS-managed llama-server:

./install.sh --no-external-llm

For another OpenAI-compatible server, such as a Lemonade Server you run yourself, use --external-llm-provider openai-compatible with the server's URL and exact model id.

This integration routes text/chat inference; it does not import or synchronize Ollama/LM Studio model files, VLMs, embedding models, or rerankers. The installer flags in this release are Linux-only. The Windows AMD llama-server.exe and macOS native llama-server keep their existing platform lifecycle.

Can I chat while models are downloading?

Yes. The installer first downloads a small bootstrap model (Qwen3.5 2B, about 1.3 GB) and starts it with a 64K context, then downloads your full model in the background. When the download is verified, ODS restarts the model server on the full model; requests in flight during that restart need a retry. ods status shows the bootstrap state while a switch is pending.

Where do I get help?

This documentation and TROUBLESHOOTING.md
Questions: GitHub Discussions
Bugs: GitHub Issues
Security problems: report them privately through Security → Report a vulnerability

Comparisons

See How It Compares in the README.

Built by Osmantic / The Collective

Navigation Menu

Repository navigation

FAQ.md

FAQ.md

ODS FAQ

Hardware

What hardware do I need?

How fast is it, and how many users can it serve?

Capabilities

What can ODS do?

Is it as good as a frontier cloud model?

Privacy & Security

Is it really private?

Can I use it with sensitive data?

What about model security?

Setup & Support

How hard is it to set up?

How do I get updates?

How do I back up and restore my data?

What are service templates?

Can ODS reuse a model already running in Ollama or LM Studio?

Can I chat while models are downloading?

Where do I get help?

Comparisons

FilesExpand file tree

FAQ.md

Latest commit

History

FAQ.md

File metadata and controls

ODS FAQ

Hardware

What hardware do I need?

How fast is it, and how many users can it serve?

Capabilities

What can ODS do?

Is it as good as a frontier cloud model?

Privacy & Security

Is it really private?

Can I use it with sensitive data?

What about model security?

Setup & Support

How hard is it to set up?

How do I get updates?

How do I back up and restore my data?

What are service templates?

Can ODS reuse a model already running in Ollama or LM Studio?

Can I chat while models are downloading?

Where do I get help?

Comparisons