Mastery Lite vs Mastery HQ
Lite gets you building.
HQ runs the whole crew.
Mastery Lite is the free edition: the pane workspace, local models and Cursor Agent, with no time limit. Mastery HQ is the full Agentic OS. Here is everything it adds, recorded in the real app.
At a glance
Where Lite stops
Lite is a complete app, not a trial. These are its limits, and in Mastery HQ none of them apply.
Agent windows at once
Lite3
HQNo cap
Workspaces
Lite1
HQOne per project, no cap
Agent backends
LiteLocal models + Cursor Agent
HQ13, from Claude Code to Kimi
Judged work
LiteOne pass, score shown up to 6/10
HQLoops until it clears your bar
Themes
Lite3 gradient + 2 animated
HQFull library + add-ons
Price
LiteFree, no time limit
HQOne-time licence
Every agent you pay for
Bring the subscriptions you already have.
Lite runs local models and Cursor Agent. Mastery HQ drives every major agent CLI side by side in the same workspace, so the Claude, ChatGPT or Gemini plan you already pay for does its work here.
- Claude Code, Codex, Gemini CLI, GitHub Copilot CLI, Grok Build CLI, OpenCode, Aider, Qwen and Kimi, plus OpenRouter and Hermes.
- As many agent windows as your machine will carry, not three.
- A separate workspace per client or project, each with its own layout and files.
Work that keeps going
Do not stop at the first judged pass.
In Lite the judge scores one pass and stops, and the score it shows tops out at 6/10. Mastery HQ scores the full range and re-runs the work against the judge until it clears the bar you set, stopping on its own when it plateaus.
- Quality loops: produce, judge, fix the biggest gap, repeat.
- Graphs and Super loops fan one goal out across several agents and send targeted fixes to whichever part fell short.
- You approve an estimate of cost and time at a gate before anything spends.
A crew, not a chat
Hand over a backlog and let the agents work it.
Put work on a sprint board and a multi-model crew plans, estimates, builds and judges every card. Or tell Hermes what you want in plain language and it routes the task to the agent with the best track record for that kind of work.
- Kanban Crew: cards your agents actually pull from, judged end to end.
- Backlog: an idea inbox, with any idea promotable to its own board.
- Agent Roster and supervisor control: one agent can spawn, message and retire the others through MCP tools.
Proof, not claims
Find out which model is actually better at your work.
Model Council puts several models on the same question and has a blind judge rank the answers. The Benchmark suite turns that into repeatable measurement on your own tasks, with two models compared side by side and the results exportable.
- One prompt, many models, one ranked verdict.
- Categorised benchmark tasks you can re-run after every model launch.
- Pick the model that earned it, not the one with the loudest launch.
The rest of the job
The tools around the code, in the same workspace.
Mastery HQ adds the panes that turn agent output into shipped work: where it went, what it cost, and what to do with it next.
- Command Center: shipped count, agent time, estimated spend and success rate, by time range.
- MCP Manager: add an MCP server once and HQ writes it into each agent CLI's own config.
- Writer, Prompt Library, Recipes, Files and deliverables, Data Lab, Recurring Reports, Device Preview and Marketing Studio.
Make, watch, listen
Media and the world, built in.
Generate brand images and videos on demand, storyboard a video scene by scene, and cut clips for every aspect ratio. Then keep live TV, music and an AI news radar on the same canvas you work on.
- Brand images, videos, reels and bumpers, saved into the project you are working in.
- Video Studio storyboards and a Clips editor for trims, reframes and captions.
- Live TV, a music player with Spotify, AI Radar, and the full animated theme library with add-ons.
Side by side
The full comparison
26 features are Mastery HQ only. Everything marked in both columns works the same in each edition.
| Feature | Mastery Lite | Mastery HQ |
|---|---|---|
| Agents | ||
| Agent windows open at once | 3 | No cap |
| Local models: Ollama, LM Studio, Jan, any OpenAI-compatible endpoint | ✓ | ✓ |
| Cursor Agent | ✓ | ✓ |
| Claude Code, Codex, Gemini CLI, GitHub Copilot CLI, Grok Build CLI, OpenCode, Aider, Qwen, Kimi | - | ✓ |
| OpenRouter models | - | ✓ |
| Agent memory, project scope and file attachments | ✓ | ✓ |
| Hermes orchestrator | - | ✓ |
| Agent Roster | - | ✓ |
| Supervisor agent control over MCP | - | ✓ |
| Quality and planning | ||
| Judged refine pass | One pass, up to 6/10 | Full 0-10 score |
| Quality loops that run until the work clears the bar | - | ✓ |
| Graphs and Super loops | - | ✓ |
| Kanban Crew sprint board | - | ✓ |
| Backlog | - | ✓ |
| Model Council | - | ✓ |
| Benchmark suite | - | ✓ |
| Workspace | ||
| Pane workspace, terminal, browser preview and code editor | ✓ | ✓ |
| Notes, whiteboard and annotated screenshots | ✓ | ✓ |
| Every output saved in the project folder | ✓ | ✓ |
| Workspaces | 1 | No cap |
| Themes | 3 gradient + 2 animated | Full library + add-ons |
| Recipes: save and reapply a layout | - | ✓ |
| Streaming Mode | ✓ | ✓ |
| Voice commands | ✓ | ✓ |
| Usage meter and Audit Log | ✓ | ✓ |
| Tools | ||
| GitHub and Google Workspace | ✓ | ✓ |
| MCP Manager | - | ✓ |
| Writer | - | ✓ |
| Prompt Library | - | ✓ |
| Files and deliverables | - | ✓ |
| Command Center | - | ✓ |
| Recurring Reports | - | ✓ |
| Data Lab | - | ✓ |
| Device Preview | - | ✓ |
| Marketing Studio | - | ✓ |
| AI Radar | - | ✓ |
| Media | ||
| Brand images, videos, reels and bumpers | - | ✓ |
| Video Studio | - | ✓ |
| Clips editor | - | ✓ |
| Live TV, and music with Spotify | - | ✓ |
Before you switch
The things people check first
Is Mastery Lite really free?
Yes. There is no trial, no time limit and no licence. The limits on this page are the whole difference, and they stay the same whether or not you register.
Do I have to uninstall Lite to use Mastery HQ?
No. They are separate apps with separate data, and they install and run side by side.
Is Mastery HQ a subscription?
No. A Founder licence is a one-time payment and every future update is included.
Does my code leave my machine in either edition?
No. Agents run locally as processes on your computer, against your folders, with your keys. Choose a local model and nothing leaves at all.
Ready for the whole crew?
Every agent, every tool, no caps.
A Founder licence is one payment, and every future update is included. Or keep building in Mastery Lite, free, for as long as you like.