macOS 14 (Sonoma) or later on Apple Silicon (M1, M2, M3, M4). RAM: 8 GB unified memory is the minimum and works fine on the cloud-only path; 16 GB+ is recommended if you run local MLX models. Disk: keep about 20 GB free. The app bundle is ~3 GB, and the rest covers your data plus any local model weights you download (roughly 2–18 GB each depending on model size). Intel Macs are not supported.
About five minutes. Download the .app, drag it to Applications, sign in with your OMEGA account, and it sets itself up automatically. The first launch downloads any local model weights you select; subsequent launches are instant. No CLI, no Docker, no Homebrew.
An iOS companion app is planned for post-launch. OMEGA itself runs on your Mac, where your agents, memory, and data live; the companion will let you check in, approve actions, and kick off tasks from your phone. It is not part of the current Early Access release. When it ships, every paid plan may pair one companion per named user under the applicable device terms.
Yes, and you never need a card to start. Core is free: real OMEGA on your Mac with free local models. When you want the full product, the $7 trial gives you 7 days of full Pro (200+ agents, 5,300+ skills, every feature). At checkout you make an explicit choice: start the trial, or subscribe now and skip it. The trial auto-converts to the $19.99/mo Pro subscription unless you cancel during the trial. Free Core plus a $7 week of Pro is the evaluation path, which is why every charge is final. Team bills per seat monthly (or annual) with no trial; Lifetime has no trial.
OMEGA was built by one engineer: Dylan, founder of Omega AI Systems in Wyoming. That is deliberate: founder-led and built in-house, so decisions ship in days instead of quarters and accountability for what ships is never outsourced. OMEGA's agents run a lot of the company's day-to-day operations, but they do not author the product; humans do. Support is handled in-house, prioritized by plan, never an outsourced help desk.
Yes. OMEGA Core is free: 1 device, 1 user, free local models via Omega-MLX, all 200+ agents, 1 workspace, 3 Automations, and 1 Project Canvas plus 1 Design Studio project, with community support. No card required, ever. Pick Core on the download page and you're in. Core is also the landing spot when you cancel a paid plan: your account downgrades to Core at the end of the billing period, and you keep OMEGA, your data, and local models.
Bring Your Own Key. You can connect your own API keys from OpenAI, Anthropic, Google, xAI, or any provider, connect Codex Subscription or Claude Subscription accounts where supported, or run local MLX. OMEGA orchestrates the agents; you control the runtime and spend. Keys and account tokens never leave your Mac.
Because OMEGA is the product. You get 200+ specialized agents, memory that carries across work, Brain knowledge graph, autonomous learning, full computer control, Automations, and optional Quantum Neural-Fractal Agentic AI™ (QNFA) for long-horizon work. Model providers sell access to models; OMEGA turns those models into useful work on your Mac.
Depends on how you use OMEGA and which runtime you choose. BYOK workloads use your provider account directly, Codex/Claude subscription-backed turns can use those subscriptions instead of API spend, and local-only MLX on Apple Silicon is $0. Budget controls cap spend per task, per day, and per month so nothing runs away.
Lifetime gives you every current paid OMEGA product feature for life. No subscription, no renewals, no future price increases: updates and new agents are included for a one-time $4,999.99. Limited launch cohorts at reduced pricing appear on the pricing page while they last. Lifetime is final at checkout: evaluate first with free Core or the $7 Pro trial, then buy once and own it.
You can cancel anytime; there are no refunds, and that is by design. You never have to pay to find out whether OMEGA fits: Core is free, and the $7 trial gives you 7 days of full Pro before the $19.99/mo subscription starts. Because evaluation is free, every charge is final. Cancelling takes effect at the end of the current billing period: you keep paid access until then, and after that your account downgrades to free Core. You keep OMEGA, your data, and free local models, and you can resubscribe anytime, so cancelling never strands you. Team bills per seat monthly (or annual) with no trial. Lifetime is final at checkout. Your statutory consumer rights are preserved.
Every plan starts with support built into the product: OMEGA diagnoses and fixes most issues in-app using your selected model (the in-app support chat plus the Diagnostics panel) at any hour, because troubleshooting is part of the product, not a staffed desk. The human layer on top: Core includes community support. Pro and Lifetime add email support, answered within one business day. Team gets priority support. Enterprise gets 24/7/365 support under its custom agreement. Support is handled in-house, never an outsourced help desk.
ChatGPT and Claude are language models with a chat window. OMEGA is a Mac app that can answer, remember, use tools, control your computer with permission, run Automations, launch agents, and keep learning from your work. You can also turn on QNFA when a thread needs deeper long-horizon focus.
Cursor and Windsurf are excellent agents inside a code editor. OMEGA's native macOS app covers chat, voice, research, automation, scheduled tasks, and computer control through the same memory and policies. OMEGA also speaks MCP, so you can connect specialized IDE tooling when you want deeper editor integration.
QNFA (Quantum Neural-Fractal Agentic AI™) is OMEGA's optional, user-toggled deep-cognition architecture. It uses quantum-inspired classical math in software on your Mac, not quantum hardware, to keep long-horizon work calibrated and self-correcting. You can toggle it per thread when you want the extra depth; OMEGA otherwise remains fully capable through Neural-Fractal Agentic AI™ (NFA).
Unlimited models. Cloud BYOK: OpenAI (GPT-5.6 Sol, GPT-5.6 Terra, GPT-5.6 Luna, GPT-5.5), Anthropic (Claude Fable 5, Opus 4.8, Sonnet 5, Haiku 4.5), Google (Gemini 3.1 Pro, Gemini 3.5 Flash), xAI (Grok 4, Grok 4 Heavy, Grok 4.5), Kimi K3, DeepSeek V4 and R1, Llama 4, OpenRouter, and any OpenAI-compatible endpoint. Subscription-backed: Codex Subscription and Claude Subscription where supported. Local: Omega-MLX, Ollama, LM Studio for zero-cost on-device inference.
Yes. Your memory, Brain, conversations, and Automations live in the local database on your Mac; they belong to you, not to a model provider. Swap Claude for GPT mid-project, drop to a free local Omega-MLX model, or plug in a new BYOK key, and every agent keeps the same context. There is no token markup and no lock-in: in OMEGA the model is a component you choose, not the product.
Yes. Code generation, code execution, GitHub repo ingestion, branch/commit/PR workflows, deploys to Docker, Vercel, Netlify, Cloudflare, or Render, and rollback. The /compare matrix shows OMEGA as supported on Code Generation and GitHub Repo Operator alongside Devin and Cursor. The difference: OMEGA covers the surrounding work too (chat, voice, macOS automation, business processes), so the same agent that writes the code can also send the deploy summary in Slack.
For most solo and small-team work, yes. OMEGA ships a full visual Design Studio: frames, layers, components, and an AI design copilot that turns your intent into a coherent, build-ready visual direction (a main frame, clear hierarchy, interaction states, and an accessible visual system) as concrete canvas edits you review and can undo from the toolbar. Design UI/UX, books, game assets, and more. And because it lives inside OMEGA, your designs flow straight into Project Canvas to get built: no handoff, no export dance.
A visual build canvas that takes an idea to a shipped outcome: an app, website, book, automation, campaign, internal tool, or research project. Describe the idea (a fragment is enough), bring in your files, local folders, git repos, or Figma links, pick the outcome, and OMEGA plans it as nodes on the canvas and builds it: real code and real files, with live preview, diffs, a terminal, and deploy. Design in Design Studio, build on Project Canvas, ship from the same app.
A living knowledge graph of your business, your work, and what actually worked. Every app you build, every line of code, every task and Automation compounds into it, and your agents read from it, so they never start from zero. OMEGA gets smarter the more you use it, and the Brain is where that intelligence accumulates. It lives in the local database on your Mac like everything else.
Yes. OMEGA ships with 200+ specialized agents across 8 domains, and the Agent Builder lets you add your own: build one manually with full control over its role, tools, and policies, or just describe what you want in plain English and OMEGA drafts the agent for you. Your custom agents use the same memory, Brain, and audit trail as the built-in ones.
Real examples from internal use: 'Process my customer support inbox each morning and draft responses.' 'Watch my Stripe webhooks and post dispute notifications.' 'Update my CRM from email threads.' 'Build a dashboard from my Postgres data.' 'Translate my Notion docs into Spanish weekly.' 'Search my Mac for invoices and put them in a spreadsheet.' OMEGA does all of these autonomously, with approvals where you want them and audit trails for everything.
Two ways. (1) Describe what you want in plain English and OMEGA generates the Automation as a visual graph, drag-and-drop, fully editable. (2) Build from scratch with 56+ pre-built nodes (browser actions, desktop control, AI prompts, conditions, loops, scheduled triggers, human approval gates, and more) plus unlimited custom nodes you save and reuse across Automations.
Three layers. (1) MCP: four tool servers ship bundled inside the app, and you can connect any of the 7,000+ servers in the open MCP ecosystem from Settings → Tools → MCP Servers. (2) Connected apps via Composio: link accounts like Gmail, Slack, or GitHub once and agents can act through them with your approval. (3) Automation nodes for browser actions, desktop control, email, HTTP requests, and database queries. If a tool speaks MCP or has an API, OMEGA can reach it.
A persistent instruction OMEGA executes in the background. 'Every weekday at 8am, check my inbox for orders, summarize them, and post to #ops in Slack.' OMEGA handles the scheduling, runs the Automation, attaches the cost/audit evidence, and pings you only on exceptions. Standing orders survive reboots and run from your Mac.
Yes. Standing orders run on-Mac as long as the Mac is awake. For 24/7 execution without keeping the Mac awake, enroll an approved VPS runner; Automations execute there under narrowed leases and synchronize status and audit evidence back to the native Mac app.
Yes. OMEGA runs locally on your Mac. Your data, conversations, API keys, and subscription connection metadata stay on your machine. We never see your tokens or store your data on our servers. Air-gapped deployment is a separately arranged commercial deployment option, not a different in-app feature tier.
Everything that is yours: conversations, memory, the Brain knowledge graph, documents, Automations, task history, API keys, and subscription connections all live in a local database on your machine. What leaves your Mac is only the model traffic you configure: cloud model calls go directly from your Mac to your provider under your key, and work routed to local Omega-MLX, Ollama, or LM Studio models never leaves the machine at all. OMEGA's servers handle licensing and billing, never your content; where phone verification is required at signup, the phone number you verify is shared with Twilio for the one-time verification code and line check, and nothing else.
Yes. Every plan can route supported work to local models such as Omega-MLX, Ollama, or LM Studio and avoid cloud model calls when configured that way. Air-gapped deployment, outbound egress lockdown, and zero-network operating procedures are separately arranged commercial deployment terms. Cloud APIs still require network access.
Encrypted at rest with a key derived from a per-machine secret stored in macOS Keychain. Keys and subscription connection metadata never appear in logs, audit trails, or task history. OMEGA only shows masked previews like 'sk-...4f9a', and no API-usage telemetry is collected.
Nothing bad, in either case. OMEGA is local-first: your data lives on your Mac, not on our servers, so cancelling a subscription downgrades your account to free Core with everything intact. The same architecture protects you from us: if Omega AI Systems ever wound down, the app on your Mac would keep working with local models, your data would stay exactly where it always was, and a Lifetime license would keep meaning what it says. You can also export everything anytime from Settings → Privacy → Export Data.
Your memory, soul, conversations, and Automations live in a Postgres database on your Mac. Uninstalling the Tauri app leaves that database in place. If you want a clean wipe, use the data export feature first (Settings → Privacy → Export Data) and then delete the application support directory. Everything is portable; you can re-import the same NDJSON export into a fresh install and pick up where you left off.
Each user gets their own profile with isolated agents, memory, Automations, BYOK API keys, and subscription connections. Free Core covers 1 device · 1 user. Early Access Pro covers 1 seat (2 devices · 1 user · one active session). Early Access Team is per seat, 2–49 seats, with shared workspaces. Enterprise (50+ seats) scales beyond that.
Shared work is shared; private context stays private. Team workspaces share projects, Automations, and the org automation library, so everyone builds on the same foundations. Each user keeps their own Brain and memory: your agents learn from your work without a teammate's context bleeding into yours. Usage, spend, and audit trails stay attributed per workspace and per user.
Yes. Per-workspace runtime settings mean per-workspace cost and usage tracking on every plan. BYOK calls, subscription-backed calls, local runs, every model call, and every Automation run are attributed to the workspace they ran in. Export usage per workspace per month for client billing or internal chargebacks.
Didn’t find your question?
Every contact-form submission is answered within one business day. No ticket maze.
Send it our way →