Agent view of layer.ai · standard home page
Layer, for agents
Layer is an AI platform for creative teams in games and entertainment. It brings 394 leading generative models — image, video, 3D, and audio — into one collaborative workspace, together with 193 task-specific tools, custom style training, and node-based workflow automation.
It is callable as well as usable: a versioned REST API and two MCP servers expose generation, editing, workflows, training, and workspace administration. Credentials are self-serve, pricing is usage-based with no seat fees, and every generation has a matching estimate operation so a run can be priced before it is executed.
What Layer does
- Generate
- Images, video, 3D, and audio from 394 models behind one API and one workspace.
- Edit
- Refine and edit existing assets, individually or in batches, rather than only generating from scratch.
- Train custom styles
- Train a style on a studio's own assets so outputs stay consistent and on-brand.
- Automate
- Compose multi-step production as node-based workflows, callable from the API and from MCP.
- Collaborate
- Workspaces, projects, and roles, with Creative Units pooled across the team.Billing is usage-based; there are no per-seat charges and no feature gates.
Catalogue
394 models and 193 tools, listed per category.
- All modelsEvery model available today (394).
- Image models130 models.
- Video models173 models.
- 3D models46 models.
- Audio models45 models.
- Tools193 task-specific creative tools.
How to call Layer
- REST API
- https://api.app.layer.ai/apiuri-path versioning; current versions v1 and v2. Description: https://layer.ai/openapi.json
- Layer MCP Server
- https://mcp.app.layer.ai/mcpstreamable-http, OAuth 2.0. Generate and edit images, video, 3D, and audio; run Blueprint workflows; train styles; assemble video timelines; upload and score files.
- Layer Management MCP Server
- https://mcp.app.layer.ai/mcp-managementstreamable-http, OAuth 2.0. Administer a workspace: projects, groups, members, roles, usage limits, and Creative Unit consumption. Kept separate so admin tools do not crowd the creative toolset.
- Authentication
- Personal Access Token (`layer_pat_`, self-serve in the app) or a workspace service account (client credentials at /api/oauth/token, scoped) for REST, or OAuth 2.0 authorization code + PKCE against https://auth.app.layer.ai/ for MCP.Dynamic client registration is supported, so an MCP client needs no pre-registered client id. Metadata: https://auth.app.layer.ai/.well-known/oauth-authorization-server
- Permissions
- GET operations require VIEWER in the target workspace. Operations that create, change, or delete state — or that start a run and therefore consume Creative Units — require at least MEMBER. Workspace and member administration requires ADMIN or OWNER.Authorization is role-based: there are no `read:*`/`write:*` API scopes, and a Personal Access Token inherits all of its creator's permissions.
- Rate limits
- 600 generations per 60s; 300 video-generations per 60sPer user. Exceeding a limit returns 429 with `Retry-After`; no `RateLimit-*` headers are emitted, so do not try to self-throttle from them.
- Async results
- Pass `webhook: {url, events?, metadata?}` on a submit request and Layer POSTs that request's terminal events to the URL, signed with Standard Webhooks headers. The submit response's `expected_events` lists which deliveries to wait for: the run's own outcome, plus a separate scoring event where the workspace scores automatically. Opted into per request; without one, poll the run on the `poll_interval_seconds` the response returns.Deliveries are verified with Standard Webhooks headers (https://layer.ai/docs/webhooks).
Getting access unattended
Full detail, including the credential-free endpoints: https://layer.ai/api/onboarding.json
- Sign-up
- Self-serve at https://layer.ai/login.No sales contact is required.
- Free tier
- Creating a Layer account is free and includes a one-off grant of Creative Units, enough to run real generations before subscribing. It is not a time-limited trial and it does not expire into a locked account. Continued generation needs a subscription (from $10/month) or a CU pack; there are no seat fees or feature gates.
- Credentials
- Create a Personal Access Token (prefixed `layer_pat_`) yourself in the app under Settings → Personal Access Tokens. No approval step and no sales contact. The token value is shown once. There is no REST endpoint that mints, lists, or revokes tokens.
- Sandbox
- There is no separate sandbox or test environment. Estimate endpoints (and their MCP equivalents) price a run without executing it, which is the supported way to rehearse a call for free.Stated as a negative on purpose: there is nothing else to look for.
- Cost control
- Every generation has a matching estimate operation. Call it and read `has_sufficient_creative_units` before executing a run.
Fetch next
Marketing and documentation pages are also published as Markdown. Request the same path with `Accept: text/markdown`, or append `.md` to the path.
- API index
application/jsonEvery fact on this page, plus the full MCP tool lists, as one JSON document. - AI catalog
application/jsonWhat Layer can be called on, and what this site can be read for. - API catalog
application/linkset+jsonRFC 9727 catalog: one anchor per API, as an RFC 9264 linkset. - MCP manifest
application/jsonBoth MCP servers with a paste-ready client config block. - Onboarding manifest
application/jsonSignup, free tier, credentials, and the endpoints that need none. - Pricing
application/jsonPlans, Creative Unit packs, and per-model generation costs. - OpenAPI description
application/jsonLayer REST API surface as OpenAPI 3.1 (base URL https://api.app.layer.ai/api). - llms.txt
text/plainIndex of every section of this site (full text: https://layer.ai/llms-full.txt). - This page as Markdown
text/markdownSame content, no markup. https://acceptmarkdown.com
Pages worth reading
Each is also available as Markdown by appending `.md`.
- PricingCreative Unit pricing, plans, and per-model cost.
- FeaturesPlatform capabilities in detail.
- EnterpriseSecurity, compliance (SOC 2 Type II), SSO/SCIM, and procurement.
- Success storiesStudio case studies with named outcomes.
- Developer docsREST API reference, MCP guides, quickstart.
- ChangelogWhat shipped, most recent first.
- BlogLonger-form writing on AI production for games.
- Contact salesFor procurement, volume, or custom terms.
Using this content
- Crawling
- Every page on this site is crawlable — nothing indexable sits behind a login or paywall, and the default robots.txt group disallows no path. Two named groups qualify that: the crawlers behind an answer engine are allowed explicitly, and the crawlers that exist only to build training corpora are disallowed. Crawl politely; this is a static site behind a CDN, so there is no published crawl-delay.Policy: https://layer.ai/robots.txt
- Content Signal
- search=yes, ai-input=yes, ai-train=noThis site may be indexed for search and used as input to an AI answer (grounding or retrieval, with attribution). Its content is not offered for model training. (https://contentsignals.org)
- Training
- Training a model on this content is not permitted. Using it as input to an AI answer — retrieval, grounding, summarising with attribution — is (`ai-input=yes`). No `noai`/`noimageai` meta directives are served; robots.txt and the Terms of Service are where this position is stated.Governing document: https://layer.ai/terms-of-service
- Attribution
- Cite the canonical page URL. Every page carries a <link rel="canonical">, and the Organization and WebSite JSON-LD nodes share stable @ids so "Layer" and "Layer AI" resolve to one entity on this domain.Cite this entity as "Layer" (also "Layer AI"), at https://layer.ai
Frequently asked questions
- What is Layer?
- Layer is the creative harness for teams that have to ship on-brand content: it brings leading generative models into one workspace, so you can generate and edit assets, collaborate, and scale production with node-based workflow automation, agents, and enterprise security. The foundation is already built, so your team differentiates on the work rather than on assembling it.
- Who is Layer built for?
- Layer is designed for creative teams of every size — from two-person teams to hundreds — and for the roles that carry production: creative producers, artists, art directors, creative strategists, marketers and growth teams who need consistent, production-ready, on-brand output.
- What can I create with Layer?
- Layer supports creating and refining multiple asset types, including images, video, 3D models, and audio, depending on the tools and models you use. Teams use them for brand assets, advertising and social creative, product and campaign visuals, and any production content a brand needs at volume.
- Does Layer include multiple AI models or just one?
- Layer positions itself as a single access point to a curated set of industry-leading AI models for creative production by producing Day 0 integrations, so teams don't have to juggle separate tools.
- Can Layer help us standardize style and keep outputs consistent?
- Yes. Layer highlights custom model training and style training so teams can teach Layer a unique style and generate consistent, on-brand visuals.
- Can we edit assets (not just generate new ones)?
- Yes. Layer includes editing and refinement capabilities to help teams iterate and polish outputs, not only generate from scratch. You can do this for individual assets as well as in batches, at scale.
- How does Layer support collaboration for teams?
- Layer emphasizes a collaborative workspace approach, and per its pricing model teams can add members without per-seat charges, sharing usage via pooled credits.
- What is Layer Workflows?
- Layer Workflows are purpose-built, agent-powered workflows that automate multi-step creative production — campaign variants, localized creative, recurring content drops — helping teams scale repeatable pipelines while keeping creatives in control.
- Is Layer enterprise-ready (security/compliance)?
- Layer highlights enterprise-grade security and compliance, including SOC 2 Type II, plus features like SSO/SCIM, role-based access controls, and audit and event logs that are fully exportable for governance and traceability.
- How does Layer pricing work?
- Layer uses a usage-based pricing approach with no seat fees, and you pay for what you create. The platform uses Creative Units (CUs) for consumption across models and tools.
- Can we start small and scale up?
- Yes. Layer offers subscription options that scale for teams of any size, with usage-based pricing that grows with production rather than with headcount — there are no per-seat fees.
- Where can I learn more or talk to sales?
- Layer provides a Contact Sales path for enterprise teams that need procurement-ready security, compliance details, or custom volume requirements.