Usage
$0 / month platform fee
Start without a commitment. Pay for the sandbox resources you allocate.
- $0.030 / vCPU-hour
- $0.010 / GiB-hour of memory
- Proposed per-second compute billing
- Same rate across consuming services
Sandbox infrastructure
Proposed launch pricing / USD
One usage model across every consuming service. No per-seat fee. These are proposed rates; public billing and checkout are not available yet.
Usage
Start without a commitment. Pay for the sandbox resources you allocate.
Build / Monthly subscription
A monthly sandbox budget alongside the coding tools you already subscribe to.
Committed
For predictable workloads. The commitment covers usage; it is not an extra platform fee.
Dedicated
Plan a dedicated pool for sustained demand across your services.
Make it concrete
Runtime is the sum across all sandboxes: 100 sandboxes × 10 hours = 1,000 hours.
$0.100 per sandbox-hour
No platform subscription added.
| Provider | vCPU-hour | GiB-hour | Monthly compute |
|---|---|---|---|
| Condensation · proposed | $0.0300 | $0.0100 | $100.00 |
| E2B ↗ | $0.050400 | $0.016200 | $165.60 |
| Daytona ↗ | $0.050400 | $0.016200 | $165.60 |
| Modal Sandboxes ↗ | $0.070956 | $0.024012 | $237.96 |
Official public rates checked September 7, 2026. E2B Pro adds $150/month; its Hobby tier has no platform fee but different limits. Daytona lists usage-based pricing. Modal’s sandbox rate uses physical cores (1 core = 2 vCPUs); we divide its CPU rate by two. Modal bills the higher of requested and actual resources; this estimate uses the requested amount. Hardware performance, limits, credits, regions, and support differ. This is a resource-price comparison, not a performance benchmark or total-bill quote.
Use Codex, Claude Code, or Cursor with your existing account. Condensation funds the sandbox resources separately.
Proposed retained storage: $0.08/GiB-month, prorated by time. Proposed internet egress: $0.05/GiB after 100 GiB per organization per month. Ingress is free. These charges are outside the calculator; committed discounts apply only to CPU and memory.
The proposed meter runs while CPU and memory are allocated, including idle and provisioning time. Compute stops once resources are released. Retained storage remains billable. Stop/resume and durable storage recovery are still planned.
Condensation is an internal preview. These plans guide the commercial model; they do not enable subscriptions, payments, service-scoped quotas, or an SLA. Rates need validation against fleet costs before paid availability.