Your Mac projects, in the cloud
A coding agent for your projects, in your pocket.
Cloud Coding copies your projects from your Mac to your own cloud VM, uncommitted changes included. Ask for a change from your iPhone, and a coding agent reads, edits, runs and previews your code there. Your Mac does not have to stay on.
How it works
- 1
Sync
The Cloud Coding app on your Mac keeps a snapshot of each project in the cloud, with your uncommitted and untracked files. It never touches your branch or your files.
- 2
Chat
Ask for a change in the iPhone app. The agent copies the latest snapshot to your VM and gets to work.
- 3
Ship
It runs your tests, opens a live preview of your app, and commits and pushes when you ask.
Made for coding away from your desk
Your own VM
Every user gets a private cloud VM. The agent runs commands, installs packages and starts dev servers there. The VM stops by itself when you are idle.
Pick your model
Choose among Claude and GPT models, and how hard they think, for the agent and for its helpers.
Continue from Claude Code
Type /cloud in Claude Code on your Mac to move the session to the cloud, and keep going from your phone.
Made for the phone
Dictate what you want, mention files with @, and get a notification when the reply is ready.
Pay for what you use
Add credits to your account, and use them for the AI in the app. Each request costs the model provider's price for its tokens, plus 10%. No subscription.
When you buy credits on the web, the card fee (2.9% + 30¢) is added to the price. Your credits do not expire.
- Claude and GPT models
- Your own cloud VM
- No monthly fee
| Model | Input | Cached input | Output |
|---|---|---|---|
| Claude Fable 5.2 | $8.80 | $0.22 | $44.00 |
| Claude Opus 5.5 | $4.40 | $0.22 | $22.00 |
| Claude Sonnet 5.5 | $2.20 | $0.11 | $11.00 |
| Claude Haiku 5.5 | $0.11 | $0.011 | $0.55 |
| GPT-6 Astra | $11.00 | $1.10 | $55.00 |
| GPT-6.1 Sol | $2.20 | $0.11 | $11.00 |
| GPT-6.1 Terra | $1.10 | $0.055 | $5.50 |
| GPT-6 Luna | $0.11 | $0.011 | $0.55 |
Thinking counts as output. Writing to the prompt cache costs a little more than input. As with the providers, a very long prompt costs more: over 100K tokens on Claude Haiku 5.5, and over 272K on GPT models.