---
title: "Rivet agentOS: Are Agents Really 250x Cheaper Than a Sandbox?"
date: "2026-09-01"
video_id: "7680436257075350814"
tiktok_url: "https://www.tiktok.com/@isaacdyor/video/7680436257075350814"
creator: "@isaacdyor"
category: "AI"
tags: ["agentos", "rivet", "ai agents", "sandbox", "webassembly"]
excerpt: "agentOS from Rivet runs coding agents inside your Node.js process. How its 250x cost and 50x speed claims compare with Rivet's own published benchmark tables."
---
agentOS is an Apache-2.0 library from Rivet that runs an agent's filesystem, shell and processes inside your existing Node.js backend instead of a separate sandbox VM. It is real and usable, but the "250x cheaper" headline is the most favorable case in Rivet's own benchmark table. For a full coding agent on AWS, Rivet reports 32x.

## What is agentOS?

A TikTok from @isaacdyor, part of his series on open source repos, says agents run 50 times faster and 250 times cheaper on agentOS than on a sandbox, and that you can attach a sandbox when you need native binaries. The code lives at github.com/rivet-dev/agentos, and the README banner reads "92x faster cold starts, 47x less memory, 254x cheaper than sandboxes."

agentOS gives each agent a lightweight virtual machine that boots inside your Node.js process. Guest JavaScript runs in V8 isolates and compiled tools run as WebAssembly, so there is no container to pull and no microVM to boot. The feature that makes it more than a code runner is bindings. Agents call functions in your backend as ordinary JavaScript calls rather than over the network, so credentials stay on the host and the agent sees only inputs and outputs. A permission layer gates filesystem, network, process and environment access, and network egress is denied by default.

## Which coding agents does it support, and how do you install it?

agentOS speaks the Agent Client Protocol over JSON-RPC on stdio, which is how it hosts existing coding agents. The registry lists Pi, Claude Code, Codex and OpenCode as built-in, plus a path for custom agents. Install it with `npm install @rivet-dev/agentos @agentos-software/pi`. You can run it locally with `npx rivetkit dev` or deploy to Rivet Cloud.

## Is agentOS 50x faster than a sandbox?

Rivet's published figure is 92x, not the 50x said in the video. Its performance page measures cold start from requesting an execution to first code running, as a median of 10,000 runs on a 12th Gen Intel i7-12700KF with Node.js v24.13.0 on Linux. agentOS lands at 4.8 ms p50, 5.6 ms p95 and 6.1 ms p99. The baseline is E2B at 440 ms p50, described as the fastest mainstream sandbox provider as of March 30, 2026.

That test machine is a desktop CPU with 64 GB of DDR4. Desktop chips usually post better single-thread numbers than shared cloud vCPUs, so treat 4.8 ms as a best case for your own deployment.

## How much memory does it save?

Memory is measured against Daytona, described as the cheapest mainstream provider on the same date. A simple shell command uses about 22 MB in agentOS against Daytona's minimum of about 1024 MB, which produces the 47x figure. A full Pi coding agent session uses about 131 MB, so the same comparison for a realistic agent workload is closer to 8x.

## Is agentOS really 250x cheaper?

Only for some workloads on some hardware. Rivet's own cost table ranges from 32x to 1,738x:

- A full coding agent on AWS ARM costs $0.00000058 per execution-second, which Rivet calls 32x cheaper than Daytona.
- The same agent on Hetzner ARM comes out at 281x.
- A simple shell command on Hetzner comes out at 1,738x.

The 254x headline sits inside that range and is not the number for running a coding agent on AWS. Rivet's page says to measure on your target hardware before quoting a cost ratio.

The cost math also assumes you provision your own servers at 70% host utilization, while the baseline runs one agent per sandbox. Sandbox providers bill per VM-second and absorb idle capacity, provisioning and operations. agentOS moves those costs onto you as servers you pay for and a utilization rate you have to hit.

## Is agentOS as secure as a sandbox?

The isolation boundary is different. A sandbox provider puts a full Linux environment, usually inside a microVM, between agent-generated code and your systems. agentOS puts the agent inside your backend process, where the boundary is V8 isolates plus WebAssembly plus the permission layer.

That boundary is widely deployed, and default-deny network egress is a sensible starting point. A separate VM still gives more distance. If code escapes the isolate, it lands in the process holding your database connections. Whether that trade works for you depends on how much you trust the code your agent writes, and the video does not raise it.

## Can you still use a sandbox with agentOS?

Yes, and this part of the video is accurate. The `@rivet-dev/agentos-sandbox` package runs a full Linux sandbox alongside the VM and mounts its filesystem into the VM as a normal directory. The agent reads and writes through the same fs APIs while the sandbox handles execution. Documented providers include E2B, Daytona, Vercel, Cloudflare, Modal, Docker and local execution.

Rivet labels sandbox mounting as beta. The use cases it names are browsers, native compilation and dev servers. If your agent shells out to a compiler or drives a headless browser, you run both systems rather than replacing one with the other.

## Who makes agentOS?

agentOS launched April 4, 2026, reached v0.2 on June 25, and added a package registry on July 6. Rivet is a Y Combinator company backed by a16z Speedrun, and agentOS sits next to its other products, Actors and Workflows. Running each VM as a Rivet Actor adds persistence, sleep and wake, and orchestration, which points the adoption path toward Rivet Cloud even though the library is self-hostable.

The video's closing line, that this or something like it is how 99% of agents will run, is a prediction. The project was five months old when the video was posted.

## Bottom line

agentOS is a real Apache-2.0 project that runs coding agents in-process with V8 and WebAssembly isolation. Rivet's own numbers put the speedup at 92x for cold starts and the cost saving at 32x for a coding agent on AWS, with larger figures reserved for trivial workloads or Hetzner hardware. You trade a VM boundary for an isolate boundary, and you take on the server costs a sandbox provider would carry.

## Sources

- [rivet-dev/agentos on GitHub](https://github.com/rivet-dev/agentos): the repo and its README benchmark banner.
- [agentOS product page](https://rivet.dev/agentos/): Rivet's overview and supported agents.
- [agentOS performance benchmarks](https://rivet.dev/agentos/docs/performance/): cold start, memory and cost tables with methodology.
- [Introducing agentOS changelog](https://rivet.dev/changelog/2026-04-04-introducing-agentos/): the April 4, 2026 launch post.
- [External Sandboxes docs](https://rivet.dev/agentos/docs/sandboxes/): how sandbox mounting works and which providers it supports.
- [agentOS Registry](https://rivet.dev/agentos/registry/): built-in agents and packages.
- [Run the Pi Coding Agent in agentOS](https://rivet.dev/agentos/docs/agents/pi/): setup for the default agent.
- [Rivet on Y Combinator](https://www.ycombinator.com/companies/rivet): company background.
- [Isaac Dyor on GitHub](https://github.com/isaacdyor): the creator's profile.


## Video description and verification notes

**agentOS is a real, Apache-2.0 library from Rivet that runs an agent's filesystem, shell, and processes inside your existing backend process, and its headline "250x cheaper" number is the most favorable case in Rivet's own benchmark table.** The video is entry five in Isaac Dyor's series on open source repos, and it makes three claims: agents run 50 times faster and 250 times cheaper than on a sandbox, and you can attach a sandbox when you need native binaries. The repo exists at github.com/rivet-dev/agentos with 4,645 stars and 251 forks as of September 19, 2026, and the README banner reads "92x faster cold starts, 47x less memory, 254x cheaper than sandboxes." Rivet's own performance documentation shows that the cost multiplier swings between 32x and 1,738x depending on which workload and which hardware you pick.

## What the library actually does

agentOS gives each agent a lightweight virtual machine that boots inside your Node.js process. Guest JavaScript runs in V8 isolates and compiled tools run as WebAssembly, so there is no container to pull and no microVM to boot. Rivet describes warm VM creation as single-digit milliseconds with each VM costing tens of megabytes.

The piece that makes it more than a code runner is bindings. Agents call functions in your backend as ordinary JavaScript calls rather than over a network hop, so credentials stay on the host and the agent sees only inputs and outputs. A permission layer gates filesystem, network, process, and environment access, with network egress denied by default.

It speaks the Agent Client Protocol over JSON-RPC on stdio, which is how it hosts existing coding agents. The registry lists Pi, Claude Code, Codex, and OpenCode as built-in, plus a path for custom agents. Installation is `npm install @rivet-dev/agentos @agentos-software/pi`, and you can run it locally with `npx rivetkit dev` or deploy to Rivet Cloud.

## The benchmark numbers, and where they come from

Rivet's performance page publishes the methodology, which is more than most projects do. Cold start is measured from requesting an execution to first code running, as a median of 10,000 runs on a 12th Gen Intel i7-12700KF with Node.js v24.13.0 on Linux. agentOS lands at 4.8 ms p50, 5.6 ms p95, and 6.1 ms p99. The comparison baseline is E2B at 440 ms p50, described as the fastest mainstream sandbox provider as of March 30, 2026. That ratio is 92x, not the 50x the voiceover states. The video's on-screen text uses the cost figure instead and reads "250x cheaper than sandboxes."

Memory is measured against Daytona, described as the cheapest mainstream provider on the same date. A simple shell command uses roughly 22 MB in agentOS versus Daytona's ~1024 MB minimum, which produces the 47x figure. A full Pi coding agent session uses roughly 131 MB, so the same comparison for a realistic agent workload is closer to 8x.

Cost follows the same pattern. For a full coding agent on AWS ARM, Rivet reports $0.00000058 per execution-second, which it calls 32x cheaper than Daytona. Move the same agent to Hetzner ARM and it reports 281x. Drop to a simple shell command on Hetzner and it reports 1,738x. The headline 254x sits inside that range, and it is not the number you get for running a coding agent on AWS. Rivet's own page says to measure on your target hardware before quoting a cost ratio.

## The comparison is partly an accounting change

The cost multipliers assume you provision your own hardware at 70% host utilization and run one agent per sandbox on the baseline side. Sandbox providers bill per VM-second and absorb the idle capacity, the provisioning, and the operational burden. agentOS moves that cost onto you, where it shows up as servers you bought and a utilization number you have to actually hit.

The benchmark hardware is also a desktop CPU, an i7-12700KF with 64 GB of DDR4, rather than a cloud instance. Desktop silicon usually posts better single-thread numbers than shared cloud vCPUs, so treat 4.8 ms as a ceiling for your own deployment rather than a figure to plan against.

## The security framing deserves a harder look

The video opens by saying that running agents in the cloud securely normally requires a sandbox, then presents agentOS as the alternative. That skips over what changes. A sandbox provider puts a full Linux environment, and usually a microVM boundary, between agent-generated code and anything you own. agentOS puts the agent inside your backend process, where the boundary is V8's isolate model plus WebAssembly plus the permission layer.

That boundary is real and widely deployed, and the default-deny network egress is a sensible posture. It is also a smaller blast radius than a separate VM, and a sandbox escape lands in the process holding your database connections. Rivet's answer is the permission system and the bindings model, which never hands the agent your credentials. Whether that trade is acceptable depends on how much you trust the code your agent generates, and the video does not raise the question.

## Sandbox mounting is the honest part

Isaac's point that you can attach a sandbox for native binaries is accurate and it is the feature that makes the project practical. agentOS ships `@rivet-dev/agentos-sandbox`, which runs a full Linux sandbox alongside the VM and projects its filesystem into the VM as a native directory. The agent reads and writes through the same fs APIs while the sandbox handles execution. Documented providers include E2B, Daytona, Vercel, Cloudflare, Modal, Docker, and local execution.

Rivet labels sandbox mounting as beta. The use cases it names are browsers, native compilation, and dev servers, which covers a lot of what people currently run coding agents for. If your agent shells out to a compiler or drives a headless browser, you are running both systems rather than replacing one with the other.

## Context the video leaves out

agentOS launched April 4, 2026, reached v0.2 on June 25, and added a package registry on July 6. Rivet is a Y Combinator company backed by a16z Speedrun, and agentOS sits next to its other products, Actors and Workflows. Running each VM as a Rivet Actor is what adds persistence, sleep and wake, and orchestration, which points the adoption path toward Rivet Cloud even though the library itself is Apache 2.0 and self-hostable.

The closing claim, that this or some version of it is how 99% of agents will be run, is a prediction and there is nothing to check it against. The project is five months old at the time of the video.

## Key Takeaways

- agentOS is a real Apache-2.0 project at github.com/rivet-dev/agentos, written primarily in Rust, with 4,645 stars and active commits as of September 19, 2026.
- The video's "250x cheaper" matches Rivet's README banner of 254x, but Rivet's own performance page reports 32x for a full coding agent on AWS ARM. The large multipliers require a trivial shell workload or Hetzner hardware.
- The voiceover's "50 times faster" does not match any published figure. Rivet claims 92x faster cold starts, 4.8 ms p50 against E2B's 440 ms.
- Benchmarks were run on a desktop i7-12700KF against baselines captured March 30, 2026, and Rivet tells readers to measure on their own hardware before quoting ratios.
- The security story changes rather than improves. Isolation moves from a separate VM to V8 isolates and WebAssembly inside your own backend process, backed by a default-deny permission layer.
- Sandbox mounting is documented and supports E2B, Daytona, Vercel, Cloudflare, Modal, and Docker, but Rivet marks it beta.
- Unverified: I could not confirm the npm download counts or that the 4,645 stars belong to agentOS alone. The GitHub repo was created February 7, 2024, more than two years before agentOS launched, which suggests the repository was repurposed, though I found no direct confirmation of that.

## Resources

- [rivet-dev/agentos on GitHub](https://github.com/rivet-dev/agentos) -- the repo the video points to, Apache 2.0, with the README benchmark banner
- [agentOS product page](https://rivet.dev/agentos/) -- Rivet's overview, positioning, and supported agents
- [agentOS performance benchmarks](https://rivet.dev/agentos/docs/performance/) -- the full cold start, memory, and cost tables with methodology and caveats
- [Introducing agentOS changelog](https://rivet.dev/changelog/2026-04-04-introducing-agentos/) -- the April 4, 2026 launch post with the original 6 ms and 32x claims
- [External Sandboxes docs](https://rivet.dev/agentos/docs/sandboxes/) -- how sandbox mounting works and which providers it supports
- [agentOS Registry](https://rivet.dev/agentos/registry/) -- the list of built-in agents and packages
- [Run the Pi Coding Agent in agentOS](https://rivet.dev/agentos/docs/agents/pi/) -- setup for the default built-in agent
- [Rivet on Y Combinator](https://www.ycombinator.com/companies/rivet) -- company background
- [Isaac Dyor on GitHub](https://github.com/isaacdyor) -- the creator's profile
