Free
$0/ month
For individuals and small teams getting started with coding agents.
- Up to 32 sandbox hours per month
- Up to 4 vCPUs / 8 GB RAM
- Up to 4 team members
- Up to 4 concurrent agents
Pricing
Simple, usage-based pricing
Pay for the agent compute you use, metered by the second and paused when idle. Bring your own model plans or API keys, invite your whole team, and scale up as you ship.
Pro
Most popular$128/ month
For teams scaling agent work with more compute and members.
- Up to
- Up to 8 vCPUs / 32 GB RAM
- Up to 32 team members
- Unlimited concurrent agents
- Find the optimal agent for your codebase via the Custom SWE-Bench
Enterprise
Custom
For organizations that need maximum scale, support, and control.
- Unlimited sandbox hours
- Up to 16 vCPUs / 64 GB RAM
- Unlimited team members
- Unlimited concurrent agents
- Deploy sandboxes in your own cloud or private infrastructure
- Uptime guarantees (SLAs)
- Dedicated support
- Optional forward-deployed engineer to transform your SDLC
Sandbox hour is defined for 1 vCPU / 2 GB RAM instance. Usage is billed per second, and instances pause automatically when idle.
Y Combinator companies and ed-tech startups get 20% off any paid plan. Get in touch.
Every plan includes
- Bring your own tokens
- Connect your Claude, ChatGPT, or Grok Build plan
- Or connect via API to Anthropic, OpenAI, Cursor, OpenRouter, Grok, OpenCode, and more
- Custom dev environments, with Docker support and fast startup
- Work across multiple repositories
- Add your entire team with secure role-based access management
- Keep your code secure with network sandboxing for agent sessions
- Live App Previews so you never have to check out a branch locally
- QA Check Agent to double-check everything with Playwright
- Agent-Guided Reviews to improve human review speed and quality
- Ship on the go with our 100% fully featured iOS app
- Desktop app (MacOS, Linux)
- Deep GitHub integration
- Deep Slack integration with image, PDF, and video support
- Meeting Assistant Agent that creates tickets as features and bugs are discussed
- Email Assistant Agent that creates tickets from and draft replies to customer requests
FAQ
Pricing questions.
What is a sandbox hour?
A sandbox hour is one hour of a 1 vCPU / 2 GB RAM cloud development environment — the isolated workspace where your agent builds, runs, and tests your app and serves live previews. Larger environments use proportionally more: a 4 vCPU / 8 GB environment consumes 4 sandbox hours per hour.
Usage is metered by the second, so you only pay for the exact compute you use — there's no hourly rounding. Sandbox hours are pooled across your whole workspace, never charged per seat.
Idle environments are paused automatically, so you only burn time while an agent is actively working or you're using a live preview.
Do I pay per seat?
No. Every plan includes team members — 4 on Free, 32 on Pro, and unlimited on Enterprise — at no extra cost. You only pay for sandbox hours, which are shared across your workspace.
Can I bring my own Claude, ChatGPT, or Grok plan?
Yes. Connect your Claude, ChatGPT, or Grok plan, or your own Anthropic, OpenAI, Amp, Cursor, Factory, OpenCode, OpenRouter, or Grok API keys. Your subscription covers model usage; Superconductor provides the environments and compute.
What happens if I run out of sandbox hours?
Your agents keep running — overage usage is billed at $30 / 100 sandbox hours beyond your plan's monthly allotment. You can also move up to a higher Pro tier at any time.
Do unused sandbox hours roll over?
Monthly sandbox hours reset at the start of each billing month and don't roll over.
What environment sizes are available?
Environments are fully configurable: choose any combination of 1 to 16 vCPUs, 2 to 64 GB of RAM, and 2 to 16 GB of disk, all in multiples of 2.
Do you offer discounts?
Yes — Y Combinator companies and ed-tech startups get 20% off any paid plan. Book a quick call or email help@superconductor.com, and we'll get you set up.
How is my code kept secure?
Every agent runs in its own isolated cloud environment. Credentials are encrypted, network access is sandboxed behind custom whitelists, and role-based access controls govern who on your team can do what. You decide exactly which repositories and resources each project can reach.
We follow industry-standard security best practices across the platform. See our security overview, or read more about network sandboxing.
Do you train models on my code?
Never. We don't use your code or data to train models, and every agent runs in an isolated, sandboxed environment with encrypted credentials and configurable network access.
How do I get started?
Sign up free — no credit card required — then connect GitHub and create your first project. Connect your existing Claude, ChatGPT, or SuperGrok plan, or add your own API keys, and you can start running agents right away.
I have more questions.
We're happy to help! Email help@superconductor.com or book a quick call, and we'll help you find the right plan.