Every model you use,
under one sky.
A premium desktop client for xAI Grok, Google Gemini and OpenAI GPT — chat, image & video, voice, and an agentic coding module, unified behind one hub. Bring your own credentials.
Context, history, projects and cost flow across every mode and every provider — no more juggling separate tools per vendor.
Live answers, with receipts.
Streaming chat that picks its provider per conversation — Grok, Gemini or GPT — with vision, document Q&A and live search, every claim backed by clickable citations and a running token & cost counter.
- ✓ Grok, Gemini or GPT, per conversation
- ✓ Live search, vision & document Q&A
- ✓ Real per-request cost & token counter
An agent you can actually trust.
A mini-IDE with a real coding agent running on Grok or GPT: sandboxed file tools, unified diff approval, checkpoints & undo, plan mode, project rules — and parallel subagent teams.
- ✓ Diff approval & 4 trust modes
- ✓ Checkpoints, undo & project rules (CAELO.md)
- ✓ Subagent teams with merge review
Create, edit, and keep it all.
Grok Imagine, Google's Gemini image models and Veo, or OpenAI GPT Image — all through one async job queue, with variations, edit/extend and a unified searchable gallery scoped to your projects.
- ✓ Text→image/video, edit, extend & variations
- ✓ The form follows the selected model's capabilities
- ✓ Unified gallery & per-job cost
Talk to Grok, hands-free.
Text-to-speech replies, dictation, a full voice conversation pipeline, and a low-latency realtime “Live” mode — your key never reaches the browser. Voice runs on xAI's speech models.
- ✓ Speak (TTS) & Transcribe (STT)
- ✓ Talk pipeline & realtime Live
- ✓ Per-session audio cost
MCP servers (local & remote), slash commands, hooks, skills, a package marketplace, headless CLI, plus ACP & LSP.
Bring your own credentials
Sign in with your xAI account, connect Google Cloud, or paste an API key. Credentials are encrypted by your operating system's credential store, sent only to the provider you selected, and never returned by the local API.
Zero telemetry
No analytics, no usage reporting, no Caelo cloud or proxy — requests go straight to the provider you picked. The backend binds to loopback only and every request is token-authenticated and fail-closed.
Open & auditable
Apache-2.0 licensed. The agent is sandboxed to your workspace; writes and shell commands require approval, with a secret-free environment.
Grab the file for your OS from the latest release, run it, then connect the provider you use — an xAI account or key, Google Cloud, or an OpenAI key.
Windows 10/11
Caelo-Setup-x.y.z.exe — x64 installer.
✅ Code-signed (Authenticode) & timestamped.
macOS · Apple Silicon
Caelo-x.y.z-arm64.dmg.
✅ Developer ID signed & notarized (Gatekeeper).
Linux · x64
.AppImage & .deb.
Unsigned — verify with the release SHA-256.
macOS is Apple Silicon only for now — Intel (x64) builds aren’t published yet.
Bring your own sky.
Download the installer for your platform, connect the provider you already pay for, and start.