Home Blog Pricing Careers Docs GitHub Slack community

Talk to Founder Dashboard →

Filesystem benchmarks: sandboxed vs. bare metal

Seven workloads, four runtimes. Most sandboxes leave 40% of your disk throughput on the table.

NKNoah Kim·MAR 21, 2026·18 MIN READ·Research

00 · Preface

Seven workloads, four runtimes. Most sandboxes leave 40% of your disk throughput on the table.

◆

DRAFT

We're actively editing this one. The public version will land next cycle.

NK

WRITTEN BYNoah KimResearch

Read next —FROM THE LOG

Mar 25, 2026Engineering

Harbor x TensorLake: Infrastructure for Agentic EvalsTensorLake is now integrated as a first-class environment provider in Harbor, enabling scalable agent evaluation with secure ephemeral MicroVMs.

Tensorlake TeamEngineering2 min Apr 3, 2026Engineering

Autoresearch on steroids with sandboxesAn LLM agent can propose incremental training-script improvements, but safely executing untrusted code requires isolated sandboxes with resource limits — that's where Tensorlake comes in. Here's the overnight hill-climb, end to end.

Tensorlake TeamEngineering9 min Apr 16, 2026Concepts

Suspend vs. snapshot: pause a sandbox, or save it for reuse?One is a pause button, the other is a save file. Same state, different question — and the answer shapes your cost model, your fan-out pattern, and which failures you can recover from.

Diptanu Gon ChoudhuryCEO / Co-founder12 min Apr 22, 2026Engineering

Tensorlake is now an official Harbor environment runtimeHarbor defines and evaluates terminal tasks. Tensorlake provides the MicroVM execution layer. Together they are a full evaluation stack for CLI agents.

Tensorlake TeamEngineering7 min

◆ THE SANDBOX DIGEST

Subscribe for release notes, benchmarks, deep dives.

One dispatch per month from the Tensorlake team — no spam.