# Sandbox cost calculator: Runtime, E2B, Daytona, Modal, Vercel and more Enter a sandbox size, run length, CPU use and runs a month to see the monthly cost on Runtime and 12 other providers. **Runtime is the cheapest for an agent that waits on a model.** 100,000 one-minute runs of a 2 vCPU, 4 GiB sandbox doing 20 CPU-seconds of work each cost $63.89 a month on Runtime, $276.00 on E2B or Daytona and $396.60 on Modal. Runtime bills the CPU your code uses, not the CPUs it holds. Rival rates checked 23 September 2026. ## How the price is worked out Every provider is priced on the same workload, at its published list rates. | Part | Runtime | Most other providers | | -------------- | --------------------------------------------------------------- | ---------------------------------------------- | | CPU | $0.025 per vCPU-hour of CPU actually used, floor 0.05 of a vCPU | Every allocated vCPU, for the whole run | | Memory | $0.0075 per GiB-hour reserved | Per GiB-hour, allocated or in use | | Plan fee | None | Some charge a monthly plan for longer sessions | | Minimum top-up | $10 of prepaid credit | Varies | A CPU-second is one core busy for one second. An agent that sends a prompt and waits for the answer is mostly idle: 20 CPU-seconds out of the 120 that 2 vCPUs could do in 60 seconds. Build steps and test suites come close to the whole 120. ## Example workloads A month of each, CPU and memory only. | Workload | Runtime | E2B | Daytona | Modal | Vercel Sandbox | | ---------------------------------------------------------------- | ------: | --------: | --------: | --------: | -------------: | | Agent runs: 2 vCPU, 4 GiB, 60 s, 20 CPU-seconds, 100,000 a month | $63.89 | $276.00 | $276.00 | $396.60 | $212.44 | | Every CPU busy: 2 vCPU, 4 GiB, 60 s, 120 CPU-seconds, 100,000 | $133.33 | $276.00 | $276.00 | $396.60 | $568.00 | | Long sessions: 4 vCPU, 8 GiB, 30 min, 720 CPU-seconds, 10,000 | $350.00 | $1,656.00 | $1,656.00 | $2,379.60 | $1,104.00 | | Small checks: 1 vCPU, 1 GiB, 10 s, 5 CPU-seconds, 1,000,000 | $55.56 | $185.00 | $185.00 | $263.80 | $236.67 | Runtime costs less than all four in every row, even with every CPU busy for the whole run. ## Why Runtime costs less - **Measured CPU.** Time your agent spends waiting on a model costs only the floor, a twentieth of a vCPU. Other providers bill every vCPU for the whole run. - **Lower rates on both meters.** $0.025 per vCPU-hour against E2B's and Daytona's $0.0504, and $0.0075 per GiB-hour of memory against their $0.0162. - **No plan fee.** Prepaid credit from $10, with no subscription, and 50 free sandbox hours to start, no card. - **Pause instead of idling.** A paused sandbox keeps its files, memory and running processes for $0.08 per GB a month and uses no CPU or memory time. ## Price your own runs exactly After a few runs on Runtime, the CLI prices the sandboxes you actually ran at another provider's rates, with the date of those rates: ```bash check runtime compare --from e2b ``` It reads your settled sandboxes from the last 30 days. The same figures come from `GET /v1/usage/compare` and the `runtime_usage_compare` MCP tool. ## Sources Rival rates checked 23 September 2026; their working and sources are on each comparison page and in the [pricing guide](/docs/pricing). Runtime's rates are in [pricing](/docs/pricing). Plan fees, storage, network, taxes, free credits and negotiated discounts are left out of every figure. - [Runtime vs E2B](/docs/e2b-alternative), [Runtime vs Daytona](/docs/daytona-alternative), [Runtime vs Modal](/docs/modal-sandbox-alternative), [Runtime vs Vercel Sandbox](/docs/vercel-sandbox-alternative) - [E2B vs Daytona](/compare/e2b-vs-daytona), [E2B vs Modal](/compare/e2b-vs-modal), [E2B alternatives](/compare/e2b-alternatives), [Modal alternatives](/compare/modal-alternatives)