Skip to content
AgentsIndex

Daytona

Isolated cloud sandboxes that start in about 90ms for running AI-generated code and agents.

Best for
AI platform teams who need isolated, fast sandboxes for agent code execution.
GitHub stars
71.5k

What we checked

Pages last read
Facts checked
2 facts

From 6 sources, read 28 October 2026

AI answers
Too few read

58 buyer-question answers read, November 2026

At a glance

Code

Last push
24 July 2026

For developers

SDKs
Python, TypeScript, Ruby, Go, Java
Runs
Cloud, Hybrid
Works with
Git, Slack, Google, GitHub, Claude, OpenCode and 2 more

Security

Compliance
SOC 2 Type II, HIPAA, GDPR

Daytona in our research

All research

Study

AI Tool Pricing Index, November 2026

Pricing

Usage-based

43 of the 361 tools we checked bill by usage.

Edition price, checked 14 November 2026.

Pricing pages of the AI tools, November 2026 editionas of n = 361 tools checked

What is Daytona?

Daytona is an isolated sandbox platform for AI platform teams that need safe execution for generated code and agent workflows. It combines Fast, Scalable, Stateful Infrastructure, Separated & Isolated Runtime Protection, Massive Parallelization, and Environment Snapshots, with SDK, RESTful API, and CLI control. Customers include LangChain, SambaNova, and Stanford's CS Department. Pricing is usage-based: vCPU $0.0504/h, GPU Nvidia H100 from $2.27/h preemptible, and Storage $0.000108/h per GiB after 5 GiB free.

The open-source repo is archived — what actually happened

The core daytonaio/daytona repository was archived by its owner on 3 October 2026 and is now read-only (confirmed via the GitHub API: "archived": true, and the archive banner on its releases page). Its README carries this notice: "As of June 2026, Daytona's core development has moved to a private codebase. This repository will receive no further updates, fixes, or releases." The last tagged release there is v0.190.0 (23 June 2026); the code stays public under its AGPL-3.0 license, usable as-is, with no further fixes or security patches. That is not the whole story, though: Daytona moved its SDKs, generated API clients, and CLI to a new repository, github.com/daytona/clients, created the same month and still active — pushed to as recently as 15 October 2026, with 0.224.x releases shipping multiple times a week. The Python SDK on PyPI and the TypeScript SDK on npm both published v0.224.1 on 14 October 2026, continuing the same version sequence the archived repo left off. In short: the open-source runtime is frozen and closed off from future development, but the hosted product and its client libraries are under active, fast-moving development — just no longer in the open.

What it's genuinely good at

Daytona's headline claim is sandbox creation in under 90ms — a vendor figure we have not independently benchmarked. What is demonstrable from the docs and dashboard: each sandbox is a fully isolated environment (dedicated kernel, filesystem, network namespace) built to run untrusted AI-generated code, with environment snapshots to save and resume state, OCI/Docker-image compatibility (bring your own image, a Dockerfile, or a docker-compose file), and built-in LSP support alongside SSH, VS Code, and web-terminal access for human debugging. It also runs virtual desktop sandboxes (Linux, macOS, Windows) for computer-use agents, and deploys across five regions (US East/West, EU Central, EU West, Asia-South) for latency-sensitive workloads. SambaNova's Chief Product & Strategy Officer is quoted citing provisioning speed at scale as the differentiator versus other sandboxing tools they evaluated — a customer testimonial, not an independent benchmark, but a specific and checkable one.

Pricing — usage-based, verified today

Pricing is metered rather than tiered, with the full ladder published on the homepage (verified 15 October 2026): compute vCPU $0.0504/h, memory $0.0162/GiB-h, storage $0.000108/GiB-h after the first 5 GiB free, and a $0.0858/vCPU-h surcharge for Windows sandboxes. GPU pricing (preemptible) runs from $0.57/h (RTX 4090) to $4.08/h (B300); Nvidia H100 is $2.27/h preemptible and RTX PRO 6000 is $1.74/h. New accounts get $200 in free compute. There's no flat subscription tier to budget against — cost tracks exactly what you provision, so estimate vCPU/GPU/storage hours rather than pick a plan.

Self-hosting and BYOC — what's actually customer-controlled

Daytona offers "customer-managed compute" (BYOC) for Enterprise customers, which runs the sandbox compute plane on the customer's own cloud. The control plane — orchestration, auth, billing — stays on Daytona's own infrastructure in every deployment mode; there is no fully self-hosted option today. The one way to run Daytona entirely outside its infrastructure is to build from the frozen AGPL-3.0 snapshot in the now-archived daytonaio/daytona repo, which receives no further fixes or security patches. For a regulated buyer asking "can we run this entirely ourselves," that's the real answer.

Compliance

Daytona runs a SafeBase-hosted trust center (trust.daytona.io) listing SOC 2 Type I and Type II, HIPAA, GDPR, and ISO/IEC 27001. The SOC 2 Type II report was issued 28 July 2026 by Insight Assurance with an unqualified opinion, covering Security, Availability, and Confidentiality criteria for the period 16 December 2025 to 15 March 2026. A Security Program whitepaper (v1.0, October 2026) is downloadable from the trust center without an access request; full SOC 2 reports and a penetration-test summary are available on request. This supports the gdpr/hipaa/soc2 tags on this listing as of today.

Customers — what's demonstrated versus quoted

The homepage names SambaNova, Writer, Martian, Handshake, Mintlify, and AfterQuery with linked case studies, and carries direct, attributed testimonial quotes from LangChain's Harrison Chase (CEO & Founder), Prosus's Paul van der Boor (VP of AI), and Sentry's Jernej Strasner (Director of Engineering), plus shorter quotes linked to Clay and n8n. These are vendor-published claims we have not independently corroborated beyond confirming the quoted individuals' stated roles are plausible company matches. A "Stanford CS Department" customer claim that appeared in this listing's earlier draft does not appear anywhere on Daytona's current site and could not be verified elsewhere — it has been dropped from this page.

How much does Daytona cost?

Prices excluding tax, as Daytona publishes them.

  • Storage: GiB

    $0.000108/h

    • Price per GiB after first 5 free
Daytona's pricing page

Pricing page, captured

daytona.io/pricing, checked

All plans and limits

PlanPriceMain limits
EnterprisePrice on requestNone stated
Pay as you go$0.0504 per vCPU hourNone stated

Read from daytona.io/pricing.

What we checked on Daytona

YesPartlyNo

What it costs to startYes

What the price counts

Priced purely on resource consumption: vCPU, memory, storage and GPU hours billed per second, with no seat fee and volume discounts for scale.

Compute, vCPU, $0.0504/h, Memory, GiB, $0.0162/h, Storage, GiB, $0.000108/h

daytona.io/pricingChecked

How you build with itPartly

Open source

The public repository stays available under the GNU AGPL v3, but the vendor stopped maintaining it in June 2026 and moved core development to a private codebase, so the running platform is no longer published.

As of June 2026, Daytona's core development has moved to a private codebase. This repository will receive no further updates, fixes, or releases.

github.com/daytonaio/daytonaChecked

Questions about Daytona

How much does Daytona cost?
Daytona costs $0.0504 per vcpu hour. From daytona.io/pricing, checked 18 November 2026.
Does Daytona have an API or an MCP server?
Daytona has an API.
Is Daytona open source, and can you host it yourself?
Part of Daytona is open source. From github.com/daytonaio/daytona, checked 28 October 2026.
What have we checked on Daytona?
2 facts, under these headings: What it costs to start, how you build with it.
How do we check Daytona?
We hold 2 facts on Daytona, quoted from 6 pages. Facts last checked 28 October 2026. Price last checked 18 November 2026. Last change found 21 March 2026.

How we check every listing

Repository

Stars
71,547
Last push
24 July 2026

Daytona price history

Plans for Daytona started at $0.0504 per vcpu hour on both of our reads, from 28 October 2026 to 18 November 2026.

PricePlanFirst seenLast seenReadingsSource
$0.0504 per vcpu hour28 October 202618 November 20262Pricing page, read 28 October 2026

From our research, read twice since 28 October 2026. A date is when we read the page, not when the price changed.

Similar in Agent tools and integrations

See all in Agent tools and integrations

Is Daytona yours?

Free, and the page then carries the verified owner mark. You confirm it with an email at daytona.io. Claim this listing

Report an error if a fact is wrong.