Your chats leave the building
Sensitive drafts, family questions, and work context sit on someone else’s servers. You rarely control retention — or training use.
Private AI for home, family, or lab
Private ChatGPT you actually control
One workspace for open models on your hardware and frontier models when you need them. Full history, accounts, and control stay on your machine.
Why this matters now
Open models are good enough for most daily work. Hosted chat still taxes your privacy, wallet, and patience.
Metered thinking
Rate limits and token bills punish everyday use. The model you need is often the one you ration.
Local UIs stop short
Many Ollama frontends are single-user chat shells. Real accounts, hybrid routing, projects, and workflow tools are missing.
Local first, not local only
Daily work stays private and free on your hardware. You only pay — and send data — when the task needs frontier capability.
Your chats stay yours
Full history lives in a local SQLite database. Local models keep everything on your network. Off-the-record mode skips saving when you want it.
Start simple, grow when ready
Start with Ollama or LM Studio. Add frontier providers or extra machines later — same dropdown.
One place for all your models
One picker plus a smart router. Local and frontier models side by side — pick yourself or let routing choose.
How it works
-
Local for daily work
Draft, summarize, brainstorm, and debug with open models on your machine — free and private.
-
Frontier when it counts
Flip to OpenAI, Claude, Grok, or other hosted models only for the hard reasoning tasks.
-
History stays in your SQLite
Tags, projects, multi-user accounts, and memory never leave your install unless you export them.
Built for real work, not just chatting
Screenshots from the actual app — the parts that set oMNI Chat apart.
Workspace
Projects, tags, and live code previews
Organize history with projects and hashtags. Generate HTML, JS, or SVG in chat and open a sandboxed live preview without leaving the conversation.
Privacy
Private by design — including off-the-record
Your full history lives in your database. Need a conversation that isn’t saved or remembered? Flip on off-the-record and it never lands in history.
Workflow
Hand work off to coding agents
Turn a chat decision into a concrete agent task. Review results in place — built for iterative coding, not just paste-and-hope snippets.
Memory
Memories you can inspect and pin
Preferences extracted from chats are visible, editable, and pin-able. You decide what the system remembers — not a black box.
Capabilities
Images, tools, and multi-user accounts
Generate images, call tools, and share one instance with family or teammates — each with their own account on the models you configured.
Image generation runs on a separate local image model — a capable GPU helps.
And the rest of the toolkit
Auto model router
Picks a fast or capable model for each question so you don’t have to think about it.
Multi-user support
Run one instance for family or a small lab. Real accounts, shared model config.
Profiles
Built-in profiles for brainstorm, draft, analyze, and code — plus custom ones via config.
Context management
Delete blocks, inline-edit answers, and keep long chats focused.
Built-in web search
Optional local SearXNG so models can fetch fresh information.
Attachments
Drop in images, PDFs, and text documents. Images go to vision models; document text is read in as context.
Voice in & out
Optional local speech-to-text to dictate, and text-to-speech to hear replies — via bundled sidecars.
Fetch any URL
Point a model at a web page or YouTube video and it pulls in the readable text or transcript.
Export chats
Download Markdown, HTML, or PDF when you need work to leave the app.
Docker or Mac app
Native Mac app today. Linux and Windows builds are next. Docker Compose is still the full-stack path.
Works with your runners
Ollama, LM Studio, llama.cpp, vLLM, LiteLLM, and anything OpenAI-compatible.
…and more. Read the full Manual
When to stay local vs go frontier
Open models are now good enough for most daily tasks.
Use frontier models only when it really counts.
| Task | Local open models | Frontier models |
|---|---|---|
| Draft emails & rewrite text | Excellent | Rarely needed |
| Explain code errors or stack traces | Great | Use if stuck |
| Coding (writing & debugging) | Good with strong local models (YMMV) | Much stronger |
| Brainstorm ideas & plans | Great | Optional |
| Summarize notes, articles, docs | Excellent | Usually overkill |
| Private or sensitive chats | Best choicestays on your machine | Only if you must |
| Hard reasoning or deep research | Good for many cases | Stronger |
Connect hosted providers (OpenAI, Claude, Grok, etc.) in the same dropdown whenever you want. Your chats stay in your database either way.
When to pick oMNI Chat
If you only want a polished Ollama UI, Open WebUI is excellent.
If you want true multi-user accounts, smart hybrid routing, projects & tags, live code previews, and inspectable memory in one private workspace — oMNI Chat is built for that.
- Family or small lab sharing one install with real accounts
- Local for daily work, frontier only when needed — same UI
- History, tags, projects, and control never leave your machine
Get oMNI Chat
Install the Mac app in one command, or run the Docker stack if you want Compose.
What you need
- A Mac with Apple Silicon (M1 or later) running macOS 13 or newer
- Optional: Ollama or LM Studio for local models
Linux and Windows are not published yet. Hardware tiers for local models are in the Manual.
Install with one command
curl -fsSL https://geekaholic.gitlab.io/omni-chat/install-macos.sh | bash
Downloads the latest DMG, copies Omni Chat.app into /Applications,
and runs xattr -dr com.apple.quarantine so Gatekeeper will launch it.
Prefer to install by hand? Open the DMG, drag Omni Chat to Applications, then run
xattr -dr com.apple.quarantine "/Applications/Omni Chat.app"
— or use System Settings → Privacy & Security → Open Anyway.
The one-liner already does this for you.
What success looks like
- Open Omni Chat from Applications (the installer launches it)
- Create your owner account
- Add a local endpoint (Ollama is the simplest) or a hosted provider key
- Pick a model in the composer and send a first message
What you need
- A Mac, Windows, or Linux computer (about 8 GB RAM for small local models; more for larger ones)
- Docker Desktop (free)
- A model runner such as Ollama or LM Studio
Important: You also need a running model endpoint (Ollama is the simplest). Hardware tiers are documented in the Manual.
Starter: Docker + Ollama
# 1. Install Docker Desktop and Ollama first
# 2. Start a small model in Ollama, e.g.:
# ollama pull gemma4:e4b
# Then run oMNI Chat:
git clone https://gitlab.com/geekaholic/omni-chat.git
cd omni-chat
cp .env.example .env
# Tell it to talk to your local Ollama
# (on Docker Desktop this hostname works)
echo 'CHAT_ENDPOINT_URL=http://host.docker.internal:11434' >> .env
docker compose up --pull always -dBarebones chat only. Pulls the image and starts the web app on port 8080.
Recommended: built-in web search
# After the Starter steps above…
# One-time secret for SearXNG
SECRET=$(openssl rand -hex 32)
echo "SEARXNG_SECRET=$SECRET" >> .env
docker compose \
-f compose.yaml \
-f compose.search.yaml \
up --pull always -dAdds a local SearXNG container so tool-using models can fetch fresh web results.
Power user: full stack
# After clone + cp .env.example .env (and Ollama / endpoint setup)…
# 1) Operator configs (edit after copy — paths, keys, models)
cp docs/endpoints.docker.example.json config/endpoints.json
cp docs/router.example.json config/router.json
cp docs/mcp.docker.example.json config/mcp.json
cp docs/agents.compose-example.json config/agents.json # Pi agent + SFTP
# 2) Env bits the overlays need
printf 'SEARXNG_SECRET=%s\n' "$(openssl rand -hex 32)" >> .env
# Edit endpoints.json for your backends; add API keys via api_key_env + .env
# 3) Agent handoff workspace (macOS Docker Desktop example)
docker build -t omni-agent-pi docker/agent-pi
mkdir -p ~/omni-agent-work
# Set managed_root in config/agents.json AND OMNI_AGENT_WORK_DIR to the same absolute path:
echo "OMNI_AGENT_WORK_DIR=$HOME/omni-agent-work" >> .env
ssh-keygen -t ed25519 -f config/sftp_host_key -N ""
# 4) Bring up search + router + fetch + aux + agents + MCP
docker compose \
-f compose.yaml \
-f compose.search.yaml \
-f compose.router.yaml \
-f compose.fetch.yaml \
-f compose.aux.yaml \
-f compose.agents.yaml \
-f compose.mcp.yaml \
up --pull always -d
Layers web search, smart routing, URL fetch, auxiliary model, coding-agent handoff, and MCP sidecars.
Copy examples from docs/ into config/, adapt them, then compose-up.
Needs more RAM/VRAM than Starter — see the
Manual
for memory tiers, Linux DOCKER_GID, and optional imagegen/TTS overlays.
What success looks like
- Open
http://localhost:8080 - Create your first user account
- Open the model dropdown in the composer
- Pick the Ollama model you started (for example
gemma4:e4b) - Ask: “Explain what this app is in simple terms.”
FAQ
Do I need an API key to start?
No. Local Ollama or LM Studio is enough for your first chats. Add hosted provider keys only when you want frontier models.
Can family or teammates share one install?
Yes. oMNI Chat supports multi-user accounts on a single instance. Everyone uses the models you configured, with their own history.
How is this different from Open WebUI?
Open WebUI is a great polished Ollama UI. oMNI Chat targets hybrid local + frontier routing, true multi-user accounts, projects/tags, live code previews, inspectable memory, and agent handoff in one private workspace.
Will my chats leave my machine?
With local models, inference stays on your network. Frontier models only receive prompts when you choose them. Chat history is stored in your local SQLite database either way. Off-the-record mode skips saving entirely.
What hardware do I need?
Any modern computer can run the app. Small open models work around 8 GB RAM; larger models need more VRAM or system memory. See the Manual for memory tiers and recommendations.
Is there a desktop app?
Yes. A native Mac app is available for Apple Silicon (macOS 13+). Install it with the one-line command in Get oMNI Chat. Linux and Windows builds are not published yet.
Why does macOS warn that the app is unidentified?
The Mac build is ad-hoc signed, so Gatekeeper quarantines a dragged-and-dropped DMG. The one-line installer clears quarantine for you. After a manual install, run xattr -dr com.apple.quarantine "/Applications/Omni Chat.app", or use System Settings → Privacy & Security → Open Anyway.
Where is the full documentation?
The online Manual covers configuration, endpoints, multi-user setup, and advanced features.
Open source. You run it.
Apache 2.0 licensed. Inspect the code, host it yourself, and keep full control of your data and models.