# How BenchFlow uses Daytona

**15k** sandboxes provisioned in a week with Daytona

As published on [daytona.io](https://www.daytona.io/customers/benchflow). Captured by usedby on 2026-10-27.

- Company: [BenchFlow](https://www.usedby.ai/companies/benchflow.md)
- Tool: [Daytona](https://www.usedby.ai/tools/daytona.md)
- Teams: Engineering

## The use case: how BenchFlow deploys Daytona

BenchFlow runs its agent skill benchmark evaluations inside isolated, stateful Daytona sandboxes, provisioning them in parallel for internal research, hackathons and competitions. Daytona replaced its self-hosted Docker-on-Kubernetes sandbox infrastructure and handles the sandbox lifecycle and teardown.

Summary written by usedby from the source page, in English. The figures are those of Daytona and BenchFlow, not ours.

> With Daytona, BenchFlow can run thousands of concurrent experiments in isolated, stateful sandboxes while preserving the reproducibility that benchmark runs depend on.

- **~3x**: “~3x reduction in monthly sandbox infrastructure costs”

> A single benchmark run can generate up to 15,000 independent experiments across models and tasks. That scale is only possible if you can provision thousands of isolated sandboxes at once. Daytona makes that straightforward.
>
> Xiangyi Li, Founder of BenchFlow

Source: [daytona.io](https://www.daytona.io/customers/benchflow), captured 2026-10-27.

## Other companies using Daytona

- [AfterQuery](https://www.usedby.ai/companies/afterquery.md) — Confirmed
- [Asteroid](https://www.usedby.ai/companies/asteroid.md) — Confirmed
- [Capy](https://www.usedby.ai/companies/capy.md) — Confirmed
- [Handshake](https://www.usedby.ai/companies/handshake.md) — Confirmed
- [Laude Institute](https://www.usedby.ai/companies/laude-institute.md) — Confirmed
- [Martian](https://www.usedby.ai/companies/martian.md) — Confirmed

## What usedby checked

We compared the story with its live page on 2026-10-27.

- Checked: the figure 15k is on the page; its label is our wording.
- Checked: the passage quoted above is copied word for word from the page, near the name of BenchFlow.
- Checked: BenchFlow uses Daytona. Confirmed line. Latest check across sources: 2026-10-27.
- Not checked: the result itself. We quote it; we did not measure it.

---
Source: https://www.usedby.ai/case-studies/benchflow-daytona · How we check: https://www.usedby.ai/methodology