Head to head · Code sandboxes · October 2026 research run

Docker Sandboxes vs Freestyle

Freestyle and Docker Sandboxes score within a point of each other on agent readiness, 62.4 (B) and 62.1 (B). Docker Sandboxes leads on agent ergonomics, security & auth and transparency & trust. Both do code sandboxes.

Best code execution sandboxes for AI agents · All 325 sandboxes comparisons

Which one, for what

Docker Sandboxes B

Best for Running coding agents such as Claude Code or Codex in a disposable cloud VM, or the same kit locally for free.

Ahead on

  • Agent ergonomics, 83 against 69
  • Security & auth, 69 against 53
  • Transparency & trust, 77 against 44

Watch for

Docker Agentic Platform, needed for every cloud interface, is experimental, and no SLA was found

Freestyle B

Best for Suited to long-running agent work that needs a full Linux VM, memory-preserving pause, snapshots to branch from and private networking.

Ahead on

  • Payments & pricing, 68 against 0

Also in its favour

  • A hosted endpoint, with nothing to install
  • Free to start without a card

Watch for

No terms of service or service agreement found on the site, and the privacy policy covers the website only

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Payments & pricing Freestyle +68

Docker Sandboxes 0

No x402, MPP or L402 (0). Compute is metered per second by size, but neither the docs, the Docker Sandboxes product page nor docker.com/pricing publishes a unit price. The plans page says usage and pricing show on the Usage & billing page after sign-in (0). Signup to Docker Agentic Platform asks for billing and payment details at checkout, with a one-time promotional credit after, so there is no card-free start for cloud sandboxes (0). A person subscribes in a browser and creates the access token (0). The free local sbx CLI is not the graded surface. 0+0+0+0 = 0.

Freestyle 68

No x402, MPP or L402 found in the docs, the OpenAPI file or the pricing page (0). Per-unit prices are public, $0.04032 a vCPU-hour, $0.0129 a GiB-hour of memory, $0.000086 a GiB-hour of storage and $0.02 a GB of data transfer (20). The Free plan costs $0 with 200 vCPU-hours and 400 GiB-hours of memory a month and needs no card (20). A person has to sign in through a browser first. After that an agent can mint its own key with freestyle tokens create --output json, so 5 of 20 (5). Pricing re-read on 4 November, unchanged. 0 + 20 + 20 + 5 = 45. Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 45 is 68.

Transparency & trust Docker Sandboxes +33

Docker Sandboxes 77

Closed software. The release repository's licence reserves all rights, and Docker's Subscription Service Agreement (26 August 2026) names Agentic Platform Services and Sandboxes (15). The privacy policy names the Agentic Platform and cloud sandboxes and says they're hosted in the United States, and the Agentic Platform FAQ gives retention for service logs (31 days), raw telemetry (12 months) and snapshots (seven days of non-use). The DPA wasn't read (20). Release notes call out breaking changes and deprecations by version and date, such as sbx ports, and the Agentic Platform docs warn that behaviour may change while it is experimental (10). CLI telemetry is disclosed with two opt-outs, and the hosting location is stated. No sub-processor list for cloud sandboxes was found (12). 15+20+10+12 = 57.

Freestyle 44

The platform is closed. The SDK and CLI are MIT on npm, and no terms of service or service agreement was found on www.freestyle.sh (/terms, /tos and /legal still answer 404 on 4 November), so the terms of use are unclear (8 of 30). The privacy policy, last updated 15 June 2025, covers the website only, and no DPA, retention period or statement on customer VM data was found (3 of 30). The 9 September 2026 changelog deprecates the Git, code execution, web deployment and scheduled trigger APIs, and the OpenAPI file marks every unversioned route deprecated, in both cases without a removal date (6 of 20). The privacy policy names Google Analytics and PostHog for the website. No sub-processor list or data location for VMs was found (3 of 20). 8 + 3 + 6 + 3 = 20.

Security & auth Docker Sandboxes +16

Docker Sandboxes 69

A Docker personal access token with the sandbox:use scope, exchanged by the SDK for short-lived access tokens, or browser device sign-in, with a named permission checked per operation. No PAT expiry or rotation guidance was read (25). Account and per-sandbox network policy with a deny-all preset, deny rules winning, and etags required on updates and deletes. Policy presets let a sandbox reach any host when allow-all or Open is chosen (14). Credentials are injected by a proxy outside the VM, so an agent inside never sees raw keys, and egress is limited to allowed destinations (12). Network policy decision logs per sandbox through /v1/policy-logs. No account audit log for cloud sandboxes found (8). SECURITY.md in docker/sbx-releases sends reports to security@docker.com with acknowledgement within 72 hours, and Docker's security.txt is valid per the docker-agent check of 1 November. Bug bounty and certifications unchecked (10). 25+14+12+8+10 = 69.

Freestyle 53

Account API keys are named, shown once, listed and revoked from the CLI, with no scopes or expiry found. Identity access tokens are separate, limited to granted VMs and named Linux users and to the exec and terminal routes, and revocable one by one, and since the 14 October changelog they take an expiration time (22 of 30). No documented option carries a secret in a URL query string. A new VM has no inbound or outbound network until firewall rules allow it, and identity tokens cannot create team resources. Deleting a VM takes one call with no confirmation step (14 of 20). Credential injection at the edge puts API keys, Postgres passwords and Git tokens on outbound requests without storing them in the VM, and forward auth keeps session cookies out of the guest (12 of 15). Audit logs are listed for Enterprise only, and the April 2026 changelog describes a logs dashboard (5 of 15). /.well-known/security.txt answers 404, and no disclosure policy, bug bounty, SOC 2 or ISO 27001 statement was found, re-read on 4 November (0 of 20). 22 + 14 + 12 + 5 + 0 = 53.

Agent ergonomics Docker Sandboxes +14

Docker Sandboxes 83

Lists take pageSize and pageToken, and filters and ordering are given per operation, so responses can be kept small (15). Pagination and a filter parameter of exact-match terms on list operations (18). Errors carry a stable code, a message and typed google.rpc.ErrorInfo details, with a documented action for each of eight codes (20). Idempotency-Key on mutations, If-Match etags on updates and deletes, and a documented recovery path for lost responses and failed waits (20). Kit launch helpers default to the small size, but the SDK is TypeScript only, with the sbx CLI and plain REST beside it (10). 15+18+20+20+10 = 83.

Freestyle 69

List calls take limit and offset, file reads take byte ranges, and paused VMs can be read without starting compute. exec-await returns whole stdout and stderr with no documented size control (15 of 25). VM lists filter by state, slug, search, snapshot, VPC and metadata, and snapshot lists by source VM and search (18 of 20). Every error has a stable code in SCREAMING_SNAKE_CASE and a message, and limit errors name the limit reached (18 of 20). No idempotency keys. A slug is unique per account and a repeated create answers 409, and the spec says which failures are safe to retry and warns against retrying an exec that returned VM_NON_RESPONSIVE (10 of 20). A create call needs only firewall, with defaults of 4 vCPU, 8 GiB and 32 GB. The only SDK is TypeScript, shipped with the CLI in one npm package (8 of 15). Per the 8 October check for the OpenAPI lines. The 14 October changelog adds vm.exec() timeouts up to 24 hours on VMs with an updated guest agent and recovery of a running command after a service interruption. 15 + 18 + 18 + 10 + 8 = 69.

Score by category

CategoryWeightDocker SandboxesFreestyleEdge
Reliability16%4547Freestyle +2
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%9086Docker Sandboxes +4
Agent ergonomics13%8369Docker Sandboxes +14
Security & auth14%6953Docker Sandboxes +16
Payments & pricing10%068Freestyle +68
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%7171even
Transparency & trust7%7744Docker Sandboxes +33
Negative events≤1500
TotalB 62.1/100B 62.4/100

Facts side by side

FactDocker SandboxesFreestyle
KindHTTP APIHTTP API
VendorDockerFreestyle Cloud Inc.
Hosted endpointno (local only)https://api.freestyle.sh/v5
TransportsHTTPHTTP
AuthOAuth or keyAPI key
PricingPay per useFreemium
x402nono
LicenceProprietary. The sbx binaries in docker/sbx-releases are under a licence that reserves all rights to Docker Inc., and cloud sandboxes run under a Docker Agentic Platform subscriptionProprietary service with no published terms of service found. The freestyle SDK and CLI package on npm is MIT
Read-only variant documentednono
llms.txtyesyes
Last release2026-11-162026-09-27
Terms last updated2026-08-26no document linked
Privacy policy last updated2026-08-26no document linked
Customer content may train modelsnot found in the textnot read yet
Terms restrict automated accessyesnot read yet
Terms restrict benchmarkingyesnot read yet
Terms or service can change without noticenot found in the textnot read yet
Arbitration or class-action waiveryesnot read yet
Popularity527 stars72k npm/wk

How the scores have moved

Verdicts

Docker Sandboxes

The cloud Sandboxes API has a public OpenAPI contract, idempotency keys, etag checks and credential injection that keeps keys out of the VM. Cloud access needs a paid Docker Agentic Platform subscription that is still experimental, and the docs publish no unit prices.

Freestyle

A public OpenAPI 3.1 file describes 88 operations with a stable error code on every failure, and per-VM identity tokens keep the team API key away from end users. No terms of service, product privacy policy, request rate limit or incident history was found, and the only SDK is TypeScript.

Before you call either

Docker Sandboxes

  1. Create the Docker personal access token with Allow Cloud Sandboxes access. Registry permissions alone don't grant the sandbox:use scope
  2. Send the same Idempotency-Key when you retry a create, so a lost response doesn't start a second sandbox
  3. Pass the etag from your latest read in If-Match. A stale etag returns 412 and a missing one 428
  4. Set --ttl and --on-timeout at creation. The default deletes sandboxes that can't be resumed after one hour
  5. Set the account network policy to deny-all and add allow rules per sandbox, since allow-all lets the sandbox reach any host

Freestyle

  1. Send a firewall object on every POST /v5/vms. It is required, and a VM with no rules has no outbound internet
  2. Ask the owner to run freestyle login in a browser once, then mint a key with freestyle tokens create --output json
  3. Do not retry an exec that answers 409 VM_NON_RESPONSIVE. The command may already have run
  4. Keep exec-await under 300 seconds on older guest agents. The 14 October changelog lets vm.exec() run up to 24 hours on VMs with an updated guest agent
  5. Call the /v5 paths. The unversioned duplicates in the OpenAPI file are marked deprecated

Questions

Which is better for AI agents, Docker Sandboxes or Freestyle?

Freestyle and Docker Sandboxes score within a point of each other on agent readiness, 62.4 (B) and 62.1 (B). Docker Sandboxes leads on agent ergonomics, security & auth and transparency & trust.

Do Docker Sandboxes and Freestyle need an API key?

Docker Sandboxes takes an API key or an OAuth sign-in. Freestyle needs an API key.

Can an agent call Docker Sandboxes and Freestyle without installing anything?

No hosted endpoint is listed for Docker Sandboxes. Freestyle has a hosted endpoint at https://api.freestyle.sh/v5.

Other comparisons with Docker Sandboxes or Freestyle

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.