World Labs World API (Marble)

# Fix list: World Labs World API (Marble)

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/world-labs, the October 2026 research run, assessed 28 October 2026. Grade D, 50.1 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on World Labs World API (Marble): work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.

## 1. Reliability, 45 out of 100, up to 11 more on the total

Why it scored 45: Hosted API reading. No status page found, and status.worldlabs.ai doesn't resolve (0). No readable incident history (5). Rate limits published with numbers, about 3 world-generation starts a minute and 60 an hour by default, about 30 and 90 a minute for approved accounts (15). The 429 guidance says to wait for Retry-After when present, back off with jitter, and poll an accepted operation rather than resubmit; no idempotency key (15). No SLA found, and the terms disclaim warranties, describing the service as is (0). Launched publicly on 21 January 2026 and not labelled beta or preview (10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 2. Payments & pricing, 20 out of 100, up to 10 more on the total

Why it scored 20: No x402, MPP or L402 (0). Credit costs per usage event are published without a login at $1.00 per 1,250 credits, $1.264 for a Marble 1.1 world from text (20). No free API tier. The Marble app's Free plan is a separate account, and API use needs a card and a $5 minimum purchase (0). A person signs in, adds a payment method, buys credits and creates the key in a browser (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):

The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).

- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.
- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login.
- 20, a free tier or trial that doesn't need a card.
- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).

Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.

Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol.

## 3. Security & auth, 48 out of 100, up to 9.1 more on the total

Why it scored 48: One kind of API key in the `WLT-Api-Key` header, created on the platform's API keys page, with more than one key per account. No scopes found, and revocation and rotation aren't described in the public docs (20). Worlds are private by default (`public` and `allow_id_access` false), but there is no read-only key, and DELETE removes a world and its assets permanently (5). Returns generated assets and an AI caption, not third-party text (10). A usage page lists usage events per request, and responses carry a request_id for support, with no audit log found (8). A responsible disclosure page with security@worldlabs.ai and a good-faith safe harbour, no bug bounty, no security.txt (404) and no SOC 2 or ISO 27001 found (5).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-security):

- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.
- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.
- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.
- 0 to 15, audit logs or per-call visibility for the operator.
- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.

Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing.

## 4. Maintenance & community, 10 out of 100, up to 7.9 more on the total

Why it scored 10: Closed-service reading. The last dated change found is the 2 April 2026 release note adding Marble 1.1 and 1.1 Plus, 209 days ago. Atlas (1 September 2026) is early access only and not on the API, and the API docs carry no dates (0). No three dated changes in the last 90 days (0). The Marble release notes are stale for the API, and support runs through a Discord help channel and support@worldlabs.ai; we didn't test whether either answers (7). The only client is worldlabs-api-python 0.1.0, unpublished on PyPI and last changed on 20 January 2026 (3). No CI configured in either worldlabsai repository (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):

- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## 5. Agent ergonomics, 70 out of 100, up to 4.9 more on the total

Why it scored 70: World responses are compact JSON with asset URLs, and splats come at 100k, 500k and full resolution so an agent can pick the size it downloads (18). `worlds:list` takes page_size, a cursor, and filters for status, model, tags, visibility and creation time, with sort order (20). Errors carry a `detail` string or per-field paths, 402 links to billing and failed operations carry a code and message, but some 400s come back as a bare Bad request (16). No idempotency key. A 429 start is documented as not accepted, and a repeated mesh export returns the cached result or the running operation at no cost (8). One required field, but the default model is the legacy marble-1.0, results are polled with no webhook, and there is no published SDK, only a Python client in a GitHub repository (8).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):

- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.

Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.

## 6. Transparency & trust, 58 out of 100, up to 3.7 more on the total

Made of editorial 40, provenance 76.

Why it scored 58: Closed service with clear terms of 21 January 2026, including output ownership and commercial use for paid and API users (15). The terms let World Labs use user content to train its models, with a prospective opt-out for paid users, and the privacy policy says training data may be kept longer. Retention is stated only as long as needed, two privacy policies with different dates are published (12 August and 12 November 2025), and the terms say API credits expire after a year while the pricing page says they don't (10). Model-name deprecation and a default-model change are announced without dates (5). A dated subprocessor list of seven vendors (12 November 2025), with no data locations (10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):

- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.
- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).
- 0 to 20, a deprecation policy or notices with dates.
- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).

The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.

Provenance checks not met in full (half of this category, computed from checked facts):

- Domain age: worldlabs.ai, registered 2023-11-22 (2 years) (7 of 15)
- Status page: not found (0 of 10)
- security.txt: not found (0 of 10)

## 7. Schema & documentation, 79 out of 100, up to 3.4 more on the total

Why it scored 79: A public OpenAPI 3.1 file covering all ten operations (25). llms.txt and a Markdown copy of every docs page (10). Operation descriptions state what each call does, what it returns and its errors, and the guides explain input types, pano detection and the two-step billing, with little on when not to use a call (13). Typed request bodies with a discriminated `world_prompt`, enums for export type and format, a seed range and documented limits such as 64-character names and 10 tags (13). curl, JavaScript and Python examples for every input type, an errors page with a status table and the 422 shape, though operation error codes aren't listed (13). Versioned paths (/marble/v1) and a model-name deprecation note, but no API changelog; the Marble release notes stop at 2 April 2026 (5).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 15, versioning and a public changelog.

Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- What happens to the World API, its prices and its terms after AMD's acquisition closes. Nothing was found on this
- Whether failed or moderated generations are charged. The pricing page doesn't say
- Whether API keys can be revoked or rotated from the platform. The platform pages render only with JavaScript and the docs don't describe it
- unchecked: whether the Discord help channel or support@worldlabs.ai answers, which wasn't tested
- The default export coordinate system. The export specs say OpenCV, the release note of 11 December 2025 says OpenGL

## Weaknesses

- No free API tier. Credits need a card and a $5 minimum purchase, and Marble app plans and credits don't carry over
- No status page, SLA or webhook found. Jobs of about 5 minutes are polled
- The terms license user content for model training by default. Paid users can opt out, prospectively only
- No dated API changelog. The last dated release note is 2 April 2026, and the Python client is unpublished and unchanged since January 2026
- AMD signed a definitive agreement to acquire World Labs on 28 September 2026. No statement on the World API's future was found

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Send `model` explicitly, such as `marble-1.1`. Omitting it selects the legacy `marble-1.0`, and that default is due to change
- Poll GET /marble/v1/operations/{operation_id} until `done` is true. A world takes about 5 minutes, and a non-null `error` carries `code` and `message`
- On 429, wait for Retry-After when present and back off with jitter. A start that returned an `operation_id` was accepted, so poll it rather than resend
- Multiply splat coordinates by `semantics_metadata.metric_scale_factor` and subtract `ground_plane_offset` from Y before measuring anything
- Check GET /marble/v1/credits before a batch. A request can overdraw the balance, and the overage is invoiced after the month ends

## When it's done

Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.