Compare

SpaceApp and eight other agent workspaces

What each tool states on its own official pages. A cell that says “not stated” means the page does not say; it does not mean the tool lacks it. Every cell names its source.

Last checked: . Prices and plans change; follow the source before you decide.

AspectSpaceAppConductorCursorGoogle AntigravityOpenAI CodexClaude CodeOrcaParadev (formerly Superset)Vibe Kanban
Form and platformsSelf-hosted server (Docker) on Linux, macOS or Windows 11; a browser on any computer, phone or tablet[21]
Desktop builds and an Android app		macOS desktop app; mobile app and cloud workspaces on Pro[1][2]macOS download, CLI, web, iOS app, Android web app; Agents Window local, cloud or SSH[3][5][7]macOS, Windows, Linux and Googlebook; app, IDE, CLI and SDK; web Remote Control[8]ChatGPT desktop app for macOS, Windows and Linux; CLI; Codex Cloud; phone app as a remote[11][13]CLI on macOS, Linux, WSL and Windows; desktop app for macOS, Windows and Ubuntu (beta); web and phone apps[14]Desktop app for macOS, Windows and Linux; iOS and Android companion; SSH remote[16]macOS; Linux AppImage (experimental); no Windows yet; iPhone and iPad monitoring app; remote hosts[17]Local web UI started with npx; Docker self-host; the page says it is sunsetting and community maintained[18][19]
Coding agents it runs16 AI coding CLIs, OpenCode first: Codex, Claude Code, Gemini, Autohand Code, Qwen Code, Kimi Code, Grok Build, DeepSeek, Cursor, GitHub Copilot, Hermes, Oh My Pi, Qoder, Muse Code, Droid[22]Claude Code, Codex, Cursor[1]Cursor Agent with models from several providers[3]Gemini models[8]OpenAI models[10]Claude models; Bedrock and other providers on the CLI[14]Any terminal CLI agent (Claude Code, Codex, OpenCode, Cursor, Copilot and more)[16]Any CLI agent, such as Claude Code, OpenCode and Codex[17]Claude Code, Codex, Gemini CLI, Copilot, Amp, Cursor, OpenCode, Droid, Qwen Code[19]
Parallel agents and isolationUp to 16 panes per room; CLI panes start in their room's own workspace folder (not a sandbox)[21]Parallel agents in isolated workspaces; cloud sandbox on paid plans[1][2]Cloud agents in isolated VMs; local worktrees[5][6]Several local agents in parallel; background subagents[8]Worktrees for independent chats; subagents[12][10]Subagents, background agents, cloud sessions; Remote Control can start each session in a worktree[14][15]One prompt fanned out to five agents, each in its own git worktree[16]One git worktree per agent; 100+ agents; scheduled automations[17]Local agents in parallel; automated git worktrees[18]
Memory across sessionsSpace Memory: records every agent session reads and writes[21]
Graph and vector search		not statednot statednot statednot statedCLAUDE.md and auto memory across sessions[14]not statednot statednot stated
Agent-blind loginsVault: agents fill logins and one-time codes without seeing themnot statedSecrets for cloud agents[5]not statednot statednot statednot statednot statednot stated
Managed browser, handed to youChrome panes with tabs and device sizes, handed to you for a sign-in (large profile)[21]not statedA cloud agent drives a desktop and browser; you can take over[5]Browser-in-the-loop agents[8]Browser and computer use in the app[11]Chrome integration[14]Embedded Chromium design mode[16]not statedBuilt-in browser to test sites[18]
Copilots beyond codeMail, Job, Marketing, Freelance, Security and Monitoring Copilotsnot statednot statednot statednot statednot statednot statednot statednot stated
VoiceLive voice that acts on panesnot statedGrok Voice from the Plus plan[4]not statednot statednot statednot statednot statednot stated
An orchestrator that asks you only for decisionsNous: it does the routine work and brings you releases, deletes, external and direction decisionsnot statednot statednot statednot statednot statednot statednot statednot stated
Several models on one questionResearch Council: models answer, score each other and agree a Top 3not stated/best-of-n runs several models, each in a worktree[6]not statednot statednot statedFan out, compare the results, merge the winner[16]not statednot stated
LicenceMIT[21]not statednot statednot statedCLI: Apache-2.0; app: not stated[13]Proprietary: all rights reserved, use under Anthropic's Commercial Terms[20]MIT[16]Elastic License 2.0 (not OSI open source)[17]Apache-2.0[19]
PriceFree to self-host; optional paid team pilot from €1,500[21]Free; Pro $50/month; Teams $60/user/month[2]Free; Plus $8; Premium $30; Super $100; Ultra $200; Teams from $30/user (per month); other models at API list price[4]Individuals free; business from $30/seat/month[9]ChatGPT Plus, Pro, Business and Enterprise[13]A Claude subscription or a Console account; Remote Control on Pro, Max, Team and Enterprise[14][15]Free[16]Free tier; prices not on the page[17]not stated
Your own accounts and keysYour own accounts and keys[21]
Space Gateway: subscriptions first, API key last		Your own subscriptions and keys[2]not statedYour Google plan[9]ChatGPT sign-in or an API key[13]Your own subscription or API key[14]Your own subscriptions[16]Your own keys; model calls are not proxied[17]Your own agent logins[19]

SpaceApp's column comes from its public repository and documentation.

What SpaceApp does not do

Get SpaceAppTry the live demo

Sources

  1. Conductor
  2. Conductor pricing
  3. Cursor
  4. Cursor pricing
  5. Cursor cloud agents
  6. Cursor worktrees
  7. Cursor Agents Window
  8. Google Antigravity
  9. Google Antigravity pricing
  10. OpenAI Codex docs
  11. OpenAI Codex app
  12. OpenAI Codex worktrees
  13. OpenAI Codex CLI
  14. Claude Code docs
  15. Claude Code Remote Control
  16. Orca
  17. Paradev (formerly Superset)
  18. Vibe Kanban
  19. Vibe Kanban repository
  20. Claude Code licence
  21. SpaceApp repository
  22. SpaceApp docs