diff --git a/.changeset/fal-openapi-queue.md b/.changeset/fal-openapi-queue.md
new file mode 100644
index 0000000..dc543ac
--- /dev/null
+++ b/.changeset/fal-openapi-queue.md
@@ -0,0 +1,5 @@
+---
+'@modelschemas/client': minor
+---
+
+`GET /v1/openapi/fal` with no selector is one queue document: `POST /{endpoint_id}` and `GET /{endpoint_id}/requests/{request_id}`, with every live model in the `endpoint_id` enum and untyped bodies. `?activity=` narrows the enum and inlines each model's own body schema. Adds `GET /v1/providers/{provider}`: one provider's status, counts, and live model counts per activity.
diff --git a/openapi.json b/openapi.json
index 6bd5615..6672f36 100644
--- a/openapi.json
+++ b/openapi.json
@@ -75,6 +75,46 @@
         }
       }
     },
+    "/v1/providers/{provider}": {
+      "get": {
+        "operationId": "getProvider",
+        "summary": "One provider: status, counts, live models per activity",
+        "description": "The same row as GET /v1/providers plus activities: live model counts keyed by activity, the values GET /v1/openapi/{provider}?activity= and GET /v1/models?activity= accept.",
+        "parameters": [
+          {
+            "name": "provider",
+            "in": "path",
+            "required": true,
+            "description": "Provider slug (see GET /v1/providers).",
+            "schema": {
+              "type": "string"
+            }
+          }
+        ],
+        "responses": {
+          "200": {
+            "description": "Provider detail.",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "$ref": "#/components/schemas/Provider"
+                }
+              }
+            }
+          },
+          "404": {
+            "description": "Error",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "$ref": "#/components/schemas/Error"
+                }
+              }
+            }
+          }
+        }
+      }
+    },
     "/v1/providers/{provider}/models": {
       "get": {
         "operationId": "listProviderModels",
@@ -500,7 +540,7 @@
       "get": {
         "operationId": "getProviderOpenApi",
         "summary": "Generation-only OpenAPI document for one provider",
-        "description": "Assembled from classified endpoints plus the provider connect profile (servers, auth, required headers). Not a raw upstream-spec mirror. Model-grained providers (today: FAL) require ?model=. A filter that matches no endpoint returns 400 spec_requires_selector; there is no cap on the number of paths in a provider-grained document. An unsynced provider returns 404 no_endpoints. ?model= pins the live model enum on provider-grained specs; it does not drop endpoints.",
+        "description": "Assembled from classified endpoints plus the provider connect profile (servers, auth, required headers). Not a raw upstream-spec mirror. FAL (model-grained, queue) is one document: POST /{endpoint_id} and GET /{endpoint_id}/requests/{request_id}. endpoint_id is the model raw id (x-modelschemas-encode raw: do not percent-encode slashes). With no selector the bodies are untyped; ?activity= narrows the enum and inlines each model's own body schema (x-modelschemas-model), up to 1024 models; ?model= is that one model's paths. A filter that matches no endpoint returns 400 spec_requires_selector; there is no cap on the number of paths in a provider-grained document. An unsynced provider returns 404 no_endpoints. ?model= pins the live model enum on provider-grained specs; it does not drop endpoints.",
         "parameters": [
           {
             "name": "provider",
@@ -514,7 +554,7 @@
           {
             "name": "model",
             "in": "query",
-            "description": "Pin to one model (raw id or slug). Required for model-grained providers.",
+            "description": "Pin to one model (raw id or slug).",
             "schema": {
               "type": "string"
             }
@@ -2453,6 +2493,106 @@
           }
         }
       },
+      "Provider": {
+        "type": "object",
+        "required": [
+          "id",
+          "displayName",
+          "status",
+          "counts",
+          "spec",
+          "activities",
+          "_links"
+        ],
+        "description": "One row of GET /v1/providers plus activities. Model counts are live rows; counts.deprecated holds the rest.",
+        "properties": {
+          "id": {
+            "type": "string"
+          },
+          "displayName": {
+            "type": "string"
+          },
+          "status": {
+            "type": "string"
+          },
+          "lastPolledAt": {
+            "type": [
+              "integer",
+              "null"
+            ]
+          },
+          "lastSyncedAt": {
+            "type": [
+              "integer",
+              "null"
+            ]
+          },
+          "counts": {
+            "type": "object",
+            "additionalProperties": {
+              "type": "integer"
+            }
+          },
+          "spec": {
+            "type": "object",
+            "required": [
+              "grain",
+              "endpointCount"
+            ],
+            "properties": {
+              "grain": {
+                "type": "string",
+                "enum": [
+                  "provider",
+                  "model"
+                ]
+              },
+              "endpointCount": {
+                "type": "integer"
+              }
+            }
+          },
+          "activities": {
+            "type": "object",
+            "description": "Live model count per activity. A key is present only when the provider has a live model with that activity.",
+            "propertyNames": {
+              "enum": [
+                "chat",
+                "image",
+                "video",
+                "audio",
+                "embeddings",
+                "moderation"
+              ]
+            },
+            "additionalProperties": {
+              "type": "integer"
+            }
+          },
+          "_links": {
+            "type": "object",
+            "required": [
+              "self",
+              "models",
+              "schemas"
+            ],
+            "properties": {
+              "self": {
+                "$ref": "#/components/schemas/HalLink"
+              },
+              "models": {
+                "$ref": "#/components/schemas/HalLink"
+              },
+              "schemas": {
+                "$ref": "#/components/schemas/HalLink"
+              },
+              "openapi": {
+                "$ref": "#/components/schemas/HalLink"
+              }
+            }
+          }
+        }
+      },
       "ModelLinks": {
         "type": "object",
         "required": [
diff --git a/packages/client/src/generated/index.ts b/packages/client/src/generated/index.ts
index 88865ef..b83e113 100644
--- a/packages/client/src/generated/index.ts
+++ b/packages/client/src/generated/index.ts
@@ -6,6 +6,7 @@ export {
   extractFalPricing,
   getActivitySchemas,
   getModel,
+  getProvider,
   getProviderOpenApi,
   getSchema,
   getServiceIndex,
@@ -62,11 +63,16 @@ export type {
   GetModelErrors,
   GetModelResponse,
   GetModelResponses,
+  GetProviderData,
+  GetProviderError,
+  GetProviderErrors,
   GetProviderOpenApiData,
   GetProviderOpenApiError,
   GetProviderOpenApiErrors,
   GetProviderOpenApiResponse,
   GetProviderOpenApiResponses,
+  GetProviderResponse,
+  GetProviderResponses,
   GetSchemaData,
   GetSchemaError,
   GetSchemaErrors,
@@ -109,6 +115,7 @@ export type {
   ModelReasoning,
   PriceBounds,
   PricingRange,
+  Provider,
   ProviderModelList,
   RateCard,
   RateCardBooleanInput,
diff --git a/packages/client/src/generated/sdk.gen.ts b/packages/client/src/generated/sdk.gen.ts
index 7f78dac..4113f90 100644
--- a/packages/client/src/generated/sdk.gen.ts
+++ b/packages/client/src/generated/sdk.gen.ts
@@ -24,9 +24,12 @@ import type {
   GetModelData,
   GetModelErrors,
   GetModelResponses,
+  GetProviderData,
+  GetProviderErrors,
   GetProviderOpenApiData,
   GetProviderOpenApiErrors,
   GetProviderOpenApiResponses,
+  GetProviderResponses,
   GetSchemaData,
   GetSchemaErrors,
   GetSchemaResponses,
@@ -109,6 +112,20 @@ export const listProviders = <ThrowOnError extends boolean = false>(
     ThrowOnError
   >({ url: '/v1/providers', ...options })
 
+/**
+ * One provider: status, counts, live models per activity
+ *
+ * The same row as GET /v1/providers plus activities: live model counts keyed by activity, the values GET /v1/openapi/{provider}?activity= and GET /v1/models?activity= accept.
+ */
+export const getProvider = <ThrowOnError extends boolean = false>(
+  options: Options<GetProviderData, ThrowOnError>,
+): RequestResult<GetProviderResponses, GetProviderErrors, ThrowOnError> =>
+  (options.client ?? client).get<
+    GetProviderResponses,
+    GetProviderErrors,
+    ThrowOnError
+  >({ url: '/v1/providers/{provider}', ...options })
+
 /**
  * List models for one provider
  */
@@ -211,7 +228,7 @@ export const getSchema = <ThrowOnError extends boolean = false>(
 /**
  * Generation-only OpenAPI document for one provider
  *
- * Assembled from classified endpoints plus the provider connect profile (servers, auth, required headers). Not a raw upstream-spec mirror. Model-grained providers (today: FAL) require ?model=. A filter that matches no endpoint returns 400 spec_requires_selector; there is no cap on the number of paths in a provider-grained document. An unsynced provider returns 404 no_endpoints. ?model= pins the live model enum on provider-grained specs; it does not drop endpoints.
+ * Assembled from classified endpoints plus the provider connect profile (servers, auth, required headers). Not a raw upstream-spec mirror. FAL (model-grained, queue) is one document: POST /{endpoint_id} and GET /{endpoint_id}/requests/{request_id}. endpoint_id is the model raw id (x-modelschemas-encode raw: do not percent-encode slashes). With no selector the bodies are untyped; ?activity= narrows the enum and inlines each model's own body schema (x-modelschemas-model), up to 1024 models; ?model= is that one model's paths. A filter that matches no endpoint returns 400 spec_requires_selector; there is no cap on the number of paths in a provider-grained document. An unsynced provider returns 404 no_endpoints. ?model= pins the live model enum on provider-grained specs; it does not drop endpoints.
  */
 export const getProviderOpenApi = <ThrowOnError extends boolean = false>(
   options: Options<GetProviderOpenApiData, ThrowOnError>,
diff --git a/packages/client/src/generated/types.gen.ts b/packages/client/src/generated/types.gen.ts
index 7cd2f1e..120b1ef 100644
--- a/packages/client/src/generated/types.gen.ts
+++ b/packages/client/src/generated/types.gen.ts
@@ -617,6 +617,36 @@ export type HalLink = {
       }
 }
 
+/**
+ * One row of GET /v1/providers plus activities. Model counts are live rows; counts.deprecated holds the rest.
+ */
+export type Provider = {
+  id: string
+  displayName: string
+  status: string
+  lastPolledAt?: number | null
+  lastSyncedAt?: number | null
+  counts: {
+    [key: string]: number
+  }
+  spec: {
+    grain: 'provider' | 'model'
+    endpointCount: number
+  }
+  /**
+   * Live model count per activity. A key is present only when the provider has a live model with that activity.
+   */
+  activities: {
+    [key: string]: number
+  }
+  _links: {
+    self: HalLink
+    models: HalLink
+    schemas: HalLink
+    openapi?: HalLink
+  }
+}
+
 export type ModelLinks = {
   provider: HalLink
   schemas: HalLink
@@ -1126,6 +1156,37 @@ export type ListProvidersResponses = {
 export type ListProvidersResponse =
   ListProvidersResponses[keyof ListProvidersResponses]
 
+export type GetProviderData = {
+  body?: never
+  path: {
+    /**
+     * Provider slug (see GET /v1/providers).
+     */
+    provider: string
+  }
+  query?: never
+  url: '/v1/providers/{provider}'
+}
+
+export type GetProviderErrors = {
+  /**
+   * Error
+   */
+  404: Error
+}
+
+export type GetProviderError = GetProviderErrors[keyof GetProviderErrors]
+
+export type GetProviderResponses = {
+  /**
+   * Provider detail.
+   */
+  200: Provider
+}
+
+export type GetProviderResponse =
+  GetProviderResponses[keyof GetProviderResponses]
+
 export type ListProviderModelsData = {
   body?: never
   path: {
@@ -1400,7 +1461,7 @@ export type GetProviderOpenApiData = {
   }
   query?: {
     /**
-     * Pin to one model (raw id or slug). Required for model-grained providers.
+     * Pin to one model (raw id or slug).
      */
     model?: string
     /**
diff --git a/skill/modelschemas/SKILL.md b/skill/modelschemas/SKILL.md
index 8e66568..90f420d 100644
--- a/skill/modelschemas/SKILL.md
+++ b/skill/modelschemas/SKILL.md
@@ -34,8 +34,13 @@ availability and schema questions.
      self-contained JSON Schema (URL-encode slashes in endpoint ids:
      `chat%2Fcompletions`). A listed model rawId also works as
      `{endpointId}` on grain=provider catalogs.
-   - `GET /v1/openapi/{provider}` — connectable generation OpenAPI
-     (FAL, currently the only model-grained provider, requires `?model=`)
+   - `GET /v1/providers/{provider}` — one provider's status, counts, and
+     live models per activity (the `?activity=` values).
+   - `GET /v1/openapi/{provider}` — connectable generation OpenAPI.
+     FAL is one queue document: the model id is the `endpoint_id` path
+     parameter (`x-modelschemas-encode: raw`, so do not percent-encode
+     slashes). Bare, the bodies are untyped; `?activity=` inlines each
+     model's own body schema; `?model=` is that one model's paths.
 3. **Authenticate for more** (5k req/h + webhooks). Easiest: the CLI —
    `modelschemas login` (agent-auth, autonomous mode, keys stored locally)
    or `modelschemas login --api-key`. Raw HTTP alternative:
@@ -140,6 +145,8 @@ spec syncs daily. Responses are JSON unless noted (`text/typescript`,
 
 1. GET /v1 — endpoint index.
 2. GET /v1/models?activity=chat — cross-provider catalog of chat models.
+   GET /v1/providers/{provider} — one provider's status, counts, and live
+   models per activity.
 3. GET /v1/schemas — system-wide index: every provider's activities and
    endpoint ids in one response (all schemas on the system).
 4. GET /v1/schemas/anthropic — activities + endpoint ids for one provider.
@@ -148,7 +155,11 @@ spec syncs daily. Responses are JSON unless noted (`text/typescript`,
    Grain=provider models also accept the raw id:
    `/v1/schemas/grok/image/grok-imagine-image-2.0?kind=input`.
 6. GET /v1/openapi/{provider} — generation-only OpenAPI (servers, auth,
-   paths). FAL requires ?model=. A filter that matches none: 400
+   paths). FAL is one queue document: the model id is the endpoint_id path
+   parameter (x-modelschemas-encode: raw — do not percent-encode slashes).
+   Bare, the bodies are untyped; ?activity= inlines each model's own body
+   schema (up to 1024 models); ?model= is that one model's paths. A filter
+   that matches none: 400
    spec_requires_selector. HTTP only (no MCP tool).
 7. POST /v1/validate {"provider","endpointId","payload"} — check a payload.
 8. POST /v1/estimate {"provider","model","request"?,"usage"?} — {amount,
diff --git a/src/routeTree.gen.ts b/src/routeTree.gen.ts
index 28683d2..c3d2add 100644
--- a/src/routeTree.gen.ts
+++ b/src/routeTree.gen.ts
@@ -43,6 +43,7 @@ import { Route as V1AgentsRegisterKeyRouteImport } from './routes/v1/agents/regi
 import { Route as V1ModelsIndexRouteImport } from './routes/v1/models/index'
 import { Route as V1OpenapiProviderRouteImport } from './routes/v1/openapi/$provider'
 import { Route as V1ProvidersIndexRouteImport } from './routes/v1/providers/index'
+import { Route as V1ProvidersProviderRouteImport } from './routes/v1/providers/$provider'
 import { Route as V1SchemasIndexRouteImport } from './routes/v1/schemas/index'
 import { Route as V1SubscriptionsIndexRouteImport } from './routes/v1/subscriptions/index'
 import { Route as V1SubscriptionsIdRouteImport } from './routes/v1/subscriptions/$id'
@@ -231,6 +232,11 @@ const V1ProvidersIndexRoute = V1ProvidersIndexRouteImport.update({
   path: '/v1/providers/',
   getParentRoute: () => rootRouteImport,
 } as any)
+const V1ProvidersProviderRoute = V1ProvidersProviderRouteImport.update({
+  id: '/v1/providers/$provider',
+  path: '/v1/providers/$provider',
+  getParentRoute: () => rootRouteImport,
+} as any)
 const V1SchemasIndexRoute = V1SchemasIndexRouteImport.update({
   id: '/v1/schemas/',
   path: '/v1/schemas/',
@@ -275,9 +281,9 @@ const V1ModelsProviderModelIdRoute = V1ModelsProviderModelIdRouteImport.update({
 } as any)
 const V1ProvidersProviderModelsRoute =
   V1ProvidersProviderModelsRouteImport.update({
-    id: '/v1/providers/$provider/models',
-    path: '/v1/providers/$provider/models',
-    getParentRoute: () => rootRouteImport,
+    id: '/models',
+    path: '/models',
+    getParentRoute: () => V1ProvidersProviderRoute,
   } as any)
 const V1SchemasProviderIndexRoute = V1SchemasProviderIndexRouteImport.update({
   id: '/v1/schemas/$provider/',
@@ -330,6 +336,7 @@ export interface FileRoutesByFullPath {
   '/v1/agents/me': typeof V1AgentsMeRoute
   '/v1/agents/register-key': typeof V1AgentsRegisterKeyRoute
   '/v1/openapi/$provider': typeof V1OpenapiProviderRoute
+  '/v1/providers/$provider': typeof V1ProvidersProviderRouteWithChildren
   '/v1/subscriptions/$id': typeof V1SubscriptionsIdRoute
   '/v1/models/': typeof V1ModelsIndexRoute
   '/v1/providers/': typeof V1ProvidersIndexRoute
@@ -378,6 +385,7 @@ export interface FileRoutesByTo {
   '/v1/agents/me': typeof V1AgentsMeRoute
   '/v1/agents/register-key': typeof V1AgentsRegisterKeyRoute
   '/v1/openapi/$provider': typeof V1OpenapiProviderRoute
+  '/v1/providers/$provider': typeof V1ProvidersProviderRouteWithChildren
   '/v1/subscriptions/$id': typeof V1SubscriptionsIdRoute
   '/v1/models': typeof V1ModelsIndexRoute
   '/v1/providers': typeof V1ProvidersIndexRoute
@@ -427,6 +435,7 @@ export interface FileRoutesById {
   '/v1/agents/me': typeof V1AgentsMeRoute
   '/v1/agents/register-key': typeof V1AgentsRegisterKeyRoute
   '/v1/openapi/$provider': typeof V1OpenapiProviderRoute
+  '/v1/providers/$provider': typeof V1ProvidersProviderRouteWithChildren
   '/v1/subscriptions/$id': typeof V1SubscriptionsIdRoute
   '/v1/models/': typeof V1ModelsIndexRoute
   '/v1/providers/': typeof V1ProvidersIndexRoute
@@ -477,6 +486,7 @@ export interface FileRouteTypes {
     | '/v1/agents/me'
     | '/v1/agents/register-key'
     | '/v1/openapi/$provider'
+    | '/v1/providers/$provider'
     | '/v1/subscriptions/$id'
     | '/v1/models/'
     | '/v1/providers/'
@@ -525,6 +535,7 @@ export interface FileRouteTypes {
     | '/v1/agents/me'
     | '/v1/agents/register-key'
     | '/v1/openapi/$provider'
+    | '/v1/providers/$provider'
     | '/v1/subscriptions/$id'
     | '/v1/models'
     | '/v1/providers'
@@ -573,6 +584,7 @@ export interface FileRouteTypes {
     | '/v1/agents/me'
     | '/v1/agents/register-key'
     | '/v1/openapi/$provider'
+    | '/v1/providers/$provider'
     | '/v1/subscriptions/$id'
     | '/v1/models/'
     | '/v1/providers/'
@@ -622,6 +634,7 @@ export interface RootRouteChildren {
   V1AgentsMeRoute: typeof V1AgentsMeRoute
   V1AgentsRegisterKeyRoute: typeof V1AgentsRegisterKeyRoute
   V1OpenapiProviderRoute: typeof V1OpenapiProviderRoute
+  V1ProvidersProviderRoute: typeof V1ProvidersProviderRouteWithChildren
   V1SubscriptionsIdRoute: typeof V1SubscriptionsIdRoute
   V1ModelsIndexRoute: typeof V1ModelsIndexRoute
   V1ProvidersIndexRoute: typeof V1ProvidersIndexRoute
@@ -632,7 +645,6 @@ export interface RootRouteChildren {
   V1AdminExtractFalRoute: typeof V1AdminExtractFalRoute
   V1AdminSyncProviderRoute: typeof V1AdminSyncProviderRoute
   V1ModelsProviderModelIdRoute: typeof V1ModelsProviderModelIdRoute
-  V1ProvidersProviderModelsRoute: typeof V1ProvidersProviderModelsRoute
   V1SchemasProviderIndexRoute: typeof V1SchemasProviderIndexRoute
   V1SchemasProviderActivitySplatRoute: typeof V1SchemasProviderActivitySplatRoute
   V1SchemasProviderActivityIndexRoute: typeof V1SchemasProviderActivityIndexRoute
@@ -878,6 +890,13 @@ declare module '@tanstack/react-router' {
       preLoaderRoute: typeof V1ProvidersIndexRouteImport
       parentRoute: typeof rootRouteImport
     }
+    '/v1/providers/$provider': {
+      id: '/v1/providers/$provider'
+      path: '/v1/providers/$provider'
+      fullPath: '/v1/providers/$provider'
+      preLoaderRoute: typeof V1ProvidersProviderRouteImport
+      parentRoute: typeof rootRouteImport
+    }
     '/v1/schemas/': {
       id: '/v1/schemas/'
       path: '/v1/schemas'
@@ -936,10 +955,10 @@ declare module '@tanstack/react-router' {
     }
     '/v1/providers/$provider/models': {
       id: '/v1/providers/$provider/models'
-      path: '/v1/providers/$provider/models'
+      path: '/models'
       fullPath: '/v1/providers/$provider/models'
       preLoaderRoute: typeof V1ProvidersProviderModelsRouteImport
-      parentRoute: typeof rootRouteImport
+      parentRoute: typeof V1ProvidersProviderRoute
     }
     '/v1/schemas/$provider/': {
       id: '/v1/schemas/$provider/'
@@ -965,6 +984,17 @@ declare module '@tanstack/react-router' {
   }
 }
 
+interface V1ProvidersProviderRouteChildren {
+  V1ProvidersProviderModelsRoute: typeof V1ProvidersProviderModelsRoute
+}
+
+const V1ProvidersProviderRouteChildren: V1ProvidersProviderRouteChildren = {
+  V1ProvidersProviderModelsRoute: V1ProvidersProviderModelsRoute,
+}
+
+const V1ProvidersProviderRouteWithChildren =
+  V1ProvidersProviderRoute._addFileChildren(V1ProvidersProviderRouteChildren)
+
 const rootRouteChildren: RootRouteChildren = {
   IndexRoute: IndexRoute,
   AccountRoute: AccountRoute,
@@ -1001,6 +1031,7 @@ const rootRouteChildren: RootRouteChildren = {
   V1AgentsMeRoute: V1AgentsMeRoute,
   V1AgentsRegisterKeyRoute: V1AgentsRegisterKeyRoute,
   V1OpenapiProviderRoute: V1OpenapiProviderRoute,
+  V1ProvidersProviderRoute: V1ProvidersProviderRouteWithChildren,
   V1SubscriptionsIdRoute: V1SubscriptionsIdRoute,
   V1ModelsIndexRoute: V1ModelsIndexRoute,
   V1ProvidersIndexRoute: V1ProvidersIndexRoute,
@@ -1011,7 +1042,6 @@ const rootRouteChildren: RootRouteChildren = {
   V1AdminExtractFalRoute: V1AdminExtractFalRoute,
   V1AdminSyncProviderRoute: V1AdminSyncProviderRoute,
   V1ModelsProviderModelIdRoute: V1ModelsProviderModelIdRoute,
-  V1ProvidersProviderModelsRoute: V1ProvidersProviderModelsRoute,
   V1SchemasProviderIndexRoute: V1SchemasProviderIndexRoute,
   V1SchemasProviderActivitySplatRoute: V1SchemasProviderActivitySplatRoute,
   V1SchemasProviderActivityIndexRoute: V1SchemasProviderActivityIndexRoute,
diff --git a/src/routes/v1/index.ts b/src/routes/v1/index.ts
index ba30777..6c5406c 100644
--- a/src/routes/v1/index.ts
+++ b/src/routes/v1/index.ts
@@ -14,6 +14,9 @@ export const serviceIndex = {
     self: halGet('/v1'),
     status: halGet('/v1/status'),
     providers: halGet('/v1/providers'),
+    provider: halGet('/v1/providers/{provider}', {
+      example: '/v1/providers/anthropic',
+    }),
     models: halGet(
       '/v1/models{?activity,provider,capability,q,provenance,pricing}',
       {
diff --git a/src/routes/v1/providers/$provider.ts b/src/routes/v1/providers/$provider.ts
new file mode 100644
index 0000000..4b43d8d
--- /dev/null
+++ b/src/routes/v1/providers/$provider.ts
@@ -0,0 +1,27 @@
+import { createFileRoute } from '@tanstack/react-router'
+import { env } from 'cloudflare:workers'
+
+import { getDb } from '#/db/index.ts'
+import { jsonError } from '#/server/admin.ts'
+import { getProviderCatalog, knownProviderIds } from '#/server/catalog.ts'
+import { cachedJson, catalogCache } from '#/server/http-cache.ts'
+
+export const Route = createFileRoute('/v1/providers/$provider')({
+  server: {
+    handlers: {
+      GET: async ({ request, params }) => {
+        const db = getDb(env, { replica: true })
+        const body = await getProviderCatalog(db, params.provider)
+        if (!body) {
+          const valid = (await knownProviderIds(db)).join(', ')
+          return jsonError(
+            404,
+            'unknown_provider',
+            `Unknown provider '${params.provider}'. Valid providers: ${valid}.`,
+          )
+        }
+        return cachedJson(request, body, catalogCache())
+      },
+    },
+  },
+})
diff --git a/src/server/catalog.ts b/src/server/catalog.ts
index 073f24d..8c8c1cd 100644
--- a/src/server/catalog.ts
+++ b/src/server/catalog.ts
@@ -64,7 +64,7 @@ function openApiLink(providerId: string) {
   if (!config?.connect) return undefined
   const grain = resolveSpecGrain(config)
   if (grain === 'model') {
-    return halGet(`/v1/openapi/${providerId}{?model}`, {
+    return halGet(`/v1/openapi/${providerId}{?model,activity}`, {
       contentType: OPENAPI_CONTENT_TYPE,
       example: `/v1/openapi/${providerId}?model=fal-ai/flux/dev`,
     })
@@ -222,26 +222,55 @@ function toApiModel(
 
 export type ApiModel = ReturnType<typeof toApiModel>
 
+function providerLinks(providerId: string) {
+  const openapi = openApiLink(providerId)
+  return {
+    self: halGet(`/v1/providers/${providerId}`),
+    models: halGet(`/v1/providers/${providerId}/models`),
+    schemas: halGet(`/v1/schemas/${providerId}`),
+    ...(openapi ? { openapi } : {}),
+  }
+}
+
 /** GET /v1/providers — status, counts, spec grain, and links. */
 export async function listProvidersCatalog(db: Db) {
   const status = await getServiceStatus(db)
   return {
-    providers: status.providers.map((p) => {
-      const openapi = openApiLink(p.id)
-      return {
-        ...p,
-        spec: specDescriptor(p.id, p.counts.endpoints),
-        _links: {
-          models: halGet(`/v1/providers/${p.id}/models`),
-          schemas: halGet(`/v1/schemas/${p.id}`),
-          ...(openapi ? { openapi } : {}),
-        },
-      }
-    }),
+    providers: status.providers.map((p) => ({
+      ...p,
+      spec: specDescriptor(p.id, p.counts.endpoints),
+      _links: providerLinks(p.id),
+    })),
     _links: { self: halGet('/v1/providers'), catalog: halGet('/v1/models') },
   }
 }
 
+/**
+ * GET /v1/providers/{provider} — one provider's status row plus live model
+ * counts per activity: the `?activity=` values `GET /v1/openapi/{provider}`
+ * and `GET /v1/models` accept. Null for an unknown provider.
+ */
+export async function getProviderCatalog(db: Db, providerId: string) {
+  const status = await getServiceStatus(db)
+  const provider = status.providers.find((p) => p.id === providerId)
+  if (!provider) return null
+  const rows = await db
+    .select({ activity: models.activity, count: sql<number>`count(*)` })
+    .from(models)
+    .where(and(eq(models.providerId, providerId), isNull(models.deprecatedAt)))
+    .groupBy(models.activity)
+  const activities: Partial<Record<Activity, number>> = {}
+  for (const row of rows) {
+    if (row.activity) activities[row.activity] = row.count
+  }
+  return {
+    ...provider,
+    spec: specDescriptor(providerId, provider.counts.endpoints),
+    activities,
+    _links: providerLinks(providerId),
+  }
+}
+
 /** GET /v1/models — the cross-provider, filterable catalog. */
 export async function listModelsCatalog(db: Db, filters: ModelFilters = {}) {
   const conditions: Array<SQL> = []
diff --git a/src/server/catalog.worker.test.ts b/src/server/catalog.worker.test.ts
index 88fbf3d..aa6307b 100644
--- a/src/server/catalog.worker.test.ts
+++ b/src/server/catalog.worker.test.ts
@@ -10,6 +10,7 @@ import { GPT_4O } from '../../packages/rate-card/src/fixtures/gpt-4o.ts'
 import { NANO_BANANA_2 } from '../../packages/rate-card/src/fixtures/nano-banana-2.ts'
 import {
   getModelDetail,
+  getProviderCatalog,
   listModelsCatalog,
   listProviderModels,
   listProvidersCatalog,
@@ -169,7 +170,7 @@ describe('listModelsCatalog filters', () => {
     )
     expect(fal?.spec.grain).toBe('model')
     expect(fal?._links.openapi).toMatchObject({
-      href: '/v1/openapi/fal{?model}',
+      href: '/v1/openapi/fal{?model,activity}',
       contentType: 'application/openapi+json',
       templated: true,
     })
@@ -313,7 +314,17 @@ describe('provider-scoped queries', () => {
     const result = await listProvidersCatalog(db)
     const alpha = result.providers.find((p) => p.id === 'cat-alpha')
     expect(alpha?._links.models.href).toBe('/v1/providers/cat-alpha/models')
+    expect(alpha?._links.self.href).toBe('/v1/providers/cat-alpha')
     expect(alpha?.counts.models).toBe(2)
+    expect(alpha).not.toHaveProperty('activities')
+  })
+
+  it('serves one provider with live model counts per activity', async () => {
+    const alpha = await getProviderCatalog(db, 'cat-alpha')
+    expect(alpha?.counts.models).toBe(2)
+    expect(alpha?.activities).toEqual({ chat: 1, image: 1 })
+    expect(alpha?._links.self.href).toBe('/v1/providers/cat-alpha')
+    expect(await getProviderCatalog(db, 'nope')).toBeNull()
   })
 })
 
diff --git a/src/server/hal.worker.test.ts b/src/server/hal.worker.test.ts
index 6670a5f..e30719e 100644
--- a/src/server/hal.worker.test.ts
+++ b/src/server/hal.worker.test.ts
@@ -8,6 +8,7 @@ import { serviceIndex } from '../routes/v1/index.ts'
 import { listChanges } from './changes-api.ts'
 import {
   getModelDetail,
+  getProviderCatalog,
   listModelsCatalog,
   listProvidersCatalog,
 } from './catalog.ts'
@@ -115,11 +116,17 @@ async function resolve(target: string | Record<string, unknown>) {
     case 'status':
       expect((await getServiceStatus(db)).providers.length).toBeGreaterThan(0)
       return
-    case 'providers':
-      expect((await listProvidersCatalog(db)).providers.length).toBeGreaterThan(
-        0,
-      )
+    case 'providers': {
+      if (rest.length === 0) {
+        expect(
+          (await listProvidersCatalog(db)).providers.length,
+        ).toBeGreaterThan(0)
+        return
+      }
+      const [provider] = rest
+      expect(await getProviderCatalog(db, provider ?? '')).not.toBeNull()
       return
+    }
     case 'models': {
       if (rest.length === 0) {
         const activity = url.searchParams.get('activity')
diff --git a/src/server/llms-txt.ts b/src/server/llms-txt.ts
index a678b95..007246f 100644
--- a/src/server/llms-txt.ts
+++ b/src/server/llms-txt.ts
@@ -48,6 +48,8 @@ spec syncs daily. Responses are JSON unless noted (\`text/typescript\`,
 
 1. GET /v1 — endpoint index.
 2. GET /v1/models?activity=chat — cross-provider catalog of chat models.
+   GET /v1/providers/{provider} — one provider's status, counts, and live
+   models per activity.
 3. GET /v1/schemas — system-wide index: every provider's activities and
    endpoint ids in one response (all schemas on the system).
 4. GET /v1/schemas/anthropic — activities + endpoint ids for one provider.
@@ -56,7 +58,11 @@ spec syncs daily. Responses are JSON unless noted (\`text/typescript\`,
    Grain=provider models also accept the raw id:
    \`/v1/schemas/grok/image/grok-imagine-image-2.0?kind=input\`.
 6. GET /v1/openapi/{provider} — generation-only OpenAPI (servers, auth,
-   paths). FAL requires ?model=. A filter that matches none: 400
+   paths). FAL is one queue document: the model id is the endpoint_id path
+   parameter (x-modelschemas-encode: raw — do not percent-encode slashes).
+   Bare, the bodies are untyped; ?activity= inlines each model's own body
+   schema (up to 1024 models); ?model= is that one model's paths. A filter
+   that matches none: 400
    spec_requires_selector. HTTP only (no MCP tool).
 7. POST /v1/validate {"provider","endpointId","payload"} — check a payload.
 8. POST /v1/estimate {"provider","model","request"?,"usage"?} — {amount,
diff --git a/src/server/openapi.ts b/src/server/openapi.ts
index 908185d..14e7b8c 100644
--- a/src/server/openapi.ts
+++ b/src/server/openapi.ts
@@ -348,6 +348,26 @@ export const openApiDocument = {
         },
       },
     },
+    '/v1/providers/{provider}': {
+      get: {
+        operationId: 'getProvider',
+        summary: 'One provider: status, counts, live models per activity',
+        description:
+          'The same row as GET /v1/providers plus activities: live model counts keyed by activity, the values GET /v1/openapi/{provider}?activity= and GET /v1/models?activity= accept.',
+        parameters: [providerParam],
+        responses: {
+          '200': {
+            description: 'Provider detail.',
+            content: {
+              'application/json': {
+                schema: { $ref: '#/components/schemas/Provider' },
+              },
+            },
+          },
+          '404': errorResponse,
+        },
+      },
+    },
     '/v1/providers/{provider}/models': {
       get: {
         operationId: 'listProviderModels',
@@ -564,7 +584,8 @@ export const openApiDocument = {
         summary: 'Generation-only OpenAPI document for one provider',
         description:
           'Assembled from classified endpoints plus the provider connect profile (servers, auth, required headers). ' +
-          'Not a raw upstream-spec mirror. Model-grained providers (today: FAL) require ?model=. ' +
+          'Not a raw upstream-spec mirror. FAL (model-grained, queue) is one document: POST /{endpoint_id} and GET /{endpoint_id}/requests/{request_id}. endpoint_id is the model raw id (x-modelschemas-encode raw: do not percent-encode slashes). ' +
+          "With no selector the bodies are untyped; ?activity= narrows the enum and inlines each model's own body schema (x-modelschemas-model), up to 1024 models; ?model= is that one model's paths. " +
           'A filter that matches no endpoint returns 400 spec_requires_selector; there is no cap on the number of paths in a provider-grained document. ' +
           'An unsynced provider returns 404 no_endpoints. ?model= pins the live model enum on provider-grained specs; it does not drop endpoints.',
         parameters: [
@@ -572,8 +593,7 @@ export const openApiDocument = {
           {
             name: 'model',
             in: 'query',
-            description:
-              'Pin to one model (raw id or slug). Required for model-grained providers.',
+            description: 'Pin to one model (raw id or slug).',
             schema: { type: 'string' },
           },
           {
@@ -1203,6 +1223,53 @@ export const openApiDocument = {
           },
         },
       },
+      Provider: {
+        type: 'object',
+        required: [
+          'id',
+          'displayName',
+          'status',
+          'counts',
+          'spec',
+          'activities',
+          '_links',
+        ],
+        description:
+          'One row of GET /v1/providers plus activities. Model counts are live rows; counts.deprecated holds the rest.',
+        properties: {
+          id: { type: 'string' },
+          displayName: { type: 'string' },
+          status: { type: 'string' },
+          lastPolledAt: { type: ['integer', 'null'] },
+          lastSyncedAt: { type: ['integer', 'null'] },
+          counts: { type: 'object', additionalProperties: { type: 'integer' } },
+          spec: {
+            type: 'object',
+            required: ['grain', 'endpointCount'],
+            properties: {
+              grain: { type: 'string', enum: ['provider', 'model'] },
+              endpointCount: { type: 'integer' },
+            },
+          },
+          activities: {
+            type: 'object',
+            description:
+              'Live model count per activity. A key is present only when the provider has a live model with that activity.',
+            propertyNames: { enum: activityEnum },
+            additionalProperties: { type: 'integer' },
+          },
+          _links: {
+            type: 'object',
+            required: ['self', 'models', 'schemas'],
+            properties: {
+              self: { $ref: '#/components/schemas/HalLink' },
+              models: { $ref: '#/components/schemas/HalLink' },
+              schemas: { $ref: '#/components/schemas/HalLink' },
+              openapi: { $ref: '#/components/schemas/HalLink' },
+            },
+          },
+        },
+      },
       ModelLinks: {
         type: 'object',
         required: ['provider', 'schemas'],
diff --git a/src/server/provider-openapi.test.ts b/src/server/provider-openapi.test.ts
index e22cd3c..64adeb4 100644
--- a/src/server/provider-openapi.test.ts
+++ b/src/server/provider-openapi.test.ts
@@ -13,7 +13,11 @@ import {
 import { falProvider } from './providers/fal.ts'
 import { anthropicProvider } from './providers/anthropic.ts'
 import { getProvider } from './providers/index.ts'
-import { resolveConnect, resolveSpecGrain } from './providers/connect.ts'
+import {
+  MAX_QUEUE_SPEC_PATHS,
+  resolveConnect,
+  resolveSpecGrain,
+} from './providers/connect.ts'
 
 describe('path + header helpers', () => {
   it('extracts braced path params', () => {
@@ -215,6 +219,7 @@ describe('connect profiles', () => {
   it('declares FAL as model-grained and Anthropic as provider-grained', () => {
     expect(resolveSpecGrain(falProvider)).toBe('model')
     expect(resolveSpecGrain(anthropicProvider)).toBe('provider')
+    expect(MAX_QUEUE_SPEC_PATHS).toBe(1024)
   })
 
   it('carries Anthropic version + x-api-key and FAL sibling GET', () => {
diff --git a/src/server/provider-openapi.ts b/src/server/provider-openapi.ts
index 73b05d8..5abf279 100644
--- a/src/server/provider-openapi.ts
+++ b/src/server/provider-openapi.ts
@@ -5,10 +5,11 @@
  * the live or pinned `model` enum, and moves output onto the sibling GET
  * when `connect.siblingGet` is set.
  */
-import { and, eq, isNull } from 'drizzle-orm'
+import { and, eq, isNull, sql } from 'drizzle-orm'
+import type { SQL } from 'drizzle-orm'
 
 import type { Db } from '#/db/index.ts'
-import { activities, endpoints, schemaVersions } from '#/db/schema.ts'
+import { activities, endpoints, models, schemaVersions } from '#/db/schema.ts'
 import type { Activity } from '#/db/schema.ts'
 import {
   getModelDetail,
@@ -18,6 +19,7 @@ import {
 import { contentHash } from '#/server/kv.ts'
 import { getProvider } from '#/server/providers/index.ts'
 import {
+  MAX_QUEUE_SPEC_PATHS,
   requiredHeaderParameters,
   resolveSpecGrain,
 } from '#/server/providers/connect.ts'
@@ -56,8 +58,12 @@ interface LoadedEndpoint {
    */
   responseContentTypes: Array<string> | null
   specRevision: string | null
+  /** Set when the endpoint was selected for a queue document. */
+  modelRawId?: string
 }
 
+const QUEUE_PATH = '/{endpoint_id}'
+
 function isObject(value: unknown): value is Record<string, unknown> {
   return typeof value === 'object' && value !== null && !Array.isArray(value)
 }
@@ -237,12 +243,12 @@ function refuseModelGrained(
   count: number,
 ): AssembleOpenApiResult {
   const example = `GET /v1/openapi/${providerId}?model=`
-  const models = `GET /v1/providers/${providerId}/models`
+  const modelList = `GET /v1/providers/${providerId}/models`
   return {
     ok: false,
     status: 400,
     code: 'spec_requires_selector',
-    message: `Provider '${providerId}' is model-grained (${String(count)} endpoints) — there is no combined OpenAPI document. Pass ?model={rawId}, e.g. ${example}<rawId>. List models: ${models}.`,
+    message: `Provider '${providerId}' is model-grained (${String(count)} endpoints) — there is no combined OpenAPI document. Pass ?model={rawId}, e.g. ${example}<rawId>. List models: ${modelList}.`,
   }
 }
 
@@ -268,10 +274,16 @@ function applyConnectProfile(
   }
 }
 
+/**
+ * `bodies: false` fetches presence only: a stored schema arrives as `{}`,
+ * so `input !== undefined` still holds without shipping and parsing every
+ * body. The lean queue document emits none of them.
+ */
 async function loadEndpoints(
   db: Db,
   providerId: string,
   activity: Activity | undefined,
+  bodies = true,
 ): Promise<Array<LoadedEndpoint>> {
   const rows = await db
     .select({
@@ -283,7 +295,11 @@ async function loadEndpoints(
       requestContentType: endpoints.requestContentType,
       responseContentTypes: endpoints.responseContentTypes,
       kind: schemaVersions.kind,
-      schema: schemaVersions.schema,
+      schema: bodies
+        ? schemaVersions.schema
+        : sql<
+            string | null
+          >`CASE WHEN ${schemaVersions.schema} IS NULL THEN NULL ELSE '{}' END`,
       specRevision: schemaVersions.specRevision,
     })
     .from(endpoints)
@@ -432,6 +448,226 @@ function buildSiblingGet(
   }
 }
 
+interface QueueModel {
+  rawId: string
+  capabilities: unknown
+}
+
+/** Live rows for a queue document, narrowed to one activity when given. */
+async function loadQueueModels(
+  db: Db,
+  providerId: string,
+  activity: Activity | undefined,
+): Promise<Array<QueueModel>> {
+  const conditions: Array<SQL> = [
+    eq(models.providerId, providerId),
+    isNull(models.deprecatedAt),
+  ]
+  if (activity) conditions.push(eq(models.activity, activity))
+  return db
+    .select({
+      rawId: models.rawId,
+      capabilities: models.capabilities,
+    })
+    .from(models)
+    .where(and(...conditions))
+    .orderBy(models.rawId)
+}
+
+/**
+ * One endpoint per queue model. The raw id is the queue path value
+ * (`https://queue.fal.run/{rawId}`), which is not the same string as the
+ * stored path (`/{rawId}`). A model with no stored endpoint is left out.
+ */
+function selectQueueEndpoints(
+  loaded: Array<LoadedEndpoint>,
+  rawIds: Array<string>,
+): Array<LoadedEndpoint> {
+  // Every key endpointMatchesModel accepts, so the match is one lookup.
+  const byKey = new Map<string, string>()
+  for (const rawId of rawIds) {
+    byKey.set(rawId, rawId)
+    byKey.set(`/${rawId}`, rawId)
+  }
+  const pending = new Set(rawIds)
+  const selected: Array<LoadedEndpoint> = []
+  for (const endpoint of loaded) {
+    const matched = byKey.get(endpoint.id) ?? byKey.get(endpoint.path)
+    if (matched === undefined || !pending.has(matched)) continue
+    pending.delete(matched)
+    selected.push({ ...endpoint, modelRawId: matched })
+  }
+  selected.sort((a, b) =>
+    (a.modelRawId ?? '').localeCompare(b.modelRawId ?? ''),
+  )
+  return selected
+}
+
+/** Component key. OpenAPI names are letters, digits, `.`, `_`, and `-`. */
+function schemaComponentKey(
+  kind: 'input' | 'output',
+  rawId: string,
+  used: Set<string>,
+): string {
+  const slug = rawId.replace(/[^A-Za-z0-9._-]+/g, '_')
+  let key = `${kind}_${slug}`
+  if (!/^[A-Za-z_]/.test(key)) key = `m_${key}`
+  key = key.slice(0, 240)
+  let unique = key
+  let n = 2
+  while (used.has(unique)) {
+    unique = `${key.slice(0, 220)}_${String(n)}`
+    n += 1
+  }
+  used.add(unique)
+  return unique
+}
+
+function withModelAnnotation(schema: unknown, rawId: string): unknown {
+  if (isObject(schema)) return { ...schema, 'x-modelschemas-model': rawId }
+  return { 'x-modelschemas-model': rawId, allOf: [schema] }
+}
+
+function schemaUnion(
+  refs: Array<Record<string, unknown>>,
+): Record<string, unknown> | undefined {
+  if (refs.length === 1) return refs[0]
+  // anyOf: FAL inputs overlap (prompt, image_url, …), so a valid body for
+  // one model often satisfies several schemas. oneOf would reject it.
+  if (refs.length > 1) return { anyOf: refs }
+  return undefined
+}
+
+/**
+ * OpenAPI path templating matches one segment and percent-encodes
+ * reserved characters. FAL ids contain slashes, so generators must append
+ * the raw id. `x-modelschemas-encode: raw` is that contract.
+ */
+function queueEndpointParameter(
+  rawIds: Array<string>,
+): Record<string, unknown> {
+  return {
+    name: 'endpoint_id',
+    in: 'path',
+    required: true,
+    'x-modelschemas-encode': 'raw',
+    description:
+      'Model raw id. Slashes are path segments: fal-ai/flux/dev is https://queue.fal.run/fal-ai/flux/dev, not fal-ai%2Fflux%2Fdev. x-modelschemas-encode is raw — do not percent-encode this value.',
+    schema: { type: 'string', enum: rawIds },
+  }
+}
+
+/**
+ * One queue client: `POST /{endpoint_id}` and
+ * `GET /{endpoint_id}/requests/{request_id}`. A model is in `endpoint_id`
+ * only when it has an input schema. The request `anyOf` is those schemas,
+ * in enum order. The result schema is included only when every included
+ * model has an output schema, in the same order. A partial set is omitted
+ * rather than filled with another model's schema.
+ *
+ * `lean` (no `?activity=`: every live model) keeps the enum and leaves the
+ * bodies untyped — a thousand inlined schemas is not a client. Each
+ * model's own schema stays at `GET /v1/schemas/{provider}/...`.
+ */
+function buildQueueDocument(
+  providerId: string,
+  selected: Array<LoadedEndpoint>,
+  connect: ProviderConnect,
+  lean: boolean,
+):
+  | {
+      paths: Record<string, Record<string, unknown>>
+      schemas: Record<string, unknown>
+    }
+  | { empty: true } {
+  const typed = selected.filter(
+    (
+      endpoint,
+    ): endpoint is LoadedEndpoint & { modelRawId: string; input: unknown } =>
+      endpoint.modelRawId !== undefined && endpoint.input !== undefined,
+  )
+  if (typed.length === 0) return { empty: true }
+  const used = new Set<string>()
+  const schemas: Record<string, unknown> = {}
+  const inputRefs: Array<Record<string, unknown>> = []
+  const outputRefs: Array<Record<string, unknown>> = []
+  const rawIds = typed.map((endpoint) => endpoint.modelRawId)
+  const outputsComplete = typed.every(
+    (endpoint) => endpoint.output !== undefined,
+  )
+  for (const endpoint of typed) {
+    if (lean) break
+    const rawId = endpoint.modelRawId
+    const inputName = schemaComponentKey('input', rawId, used)
+    schemas[inputName] = withModelAnnotation(endpoint.input, rawId)
+    inputRefs.push({ $ref: `#/components/schemas/${inputName}` })
+    if (!outputsComplete || endpoint.output === undefined) continue
+    const outputName = schemaComponentKey('output', rawId, used)
+    schemas[outputName] = withModelAnnotation(endpoint.output, rawId)
+    outputRefs.push({ $ref: `#/components/schemas/${outputName}` })
+  }
+  const activitiesPresent = new Set(typed.map((endpoint) => endpoint.activity))
+  const onlyActivity =
+    activitiesPresent.size === 1 ? typed[0]?.activity : undefined
+  const endpointParam = queueEndpointParameter(rawIds)
+  const input = lean
+    ? {
+        type: 'object',
+        description: `The model's own input schema: GET /v1/schemas/${providerId}/{activity}/{endpoint_id}?kind=input. Pass ?activity= on this document to inline them.`,
+      }
+    : schemaUnion(inputRefs)
+  const output = !lean && outputsComplete ? schemaUnion(outputRefs) : undefined
+  const post: Record<string, unknown> = {
+    operationId: operationId(providerId, 'post', QUEUE_PATH),
+    summary: onlyActivity
+      ? `Queue a ${onlyActivity} generation`
+      : 'Queue a generation',
+    ...(onlyActivity ? { 'x-modelschemas-activity': onlyActivity } : {}),
+    parameters: [endpointParam, ...headerParams(connect.requiredHeaders)],
+    ...(input
+      ? { requestBody: { required: true, ...requestMedia(input, null) } }
+      : {}),
+    responses: {
+      '200': {
+        description: 'Success',
+        ...(connect.siblingGet ? {} : (responseMedia(output, null) ?? {})),
+      },
+    },
+  }
+  const paths: Record<string, Record<string, unknown>> = {
+    [QUEUE_PATH]: { post },
+  }
+  if (connect.siblingGet) {
+    const requestPath = siblingRequestPath(QUEUE_PATH)
+    paths[requestPath] = {
+      get: {
+        operationId: operationId(providerId, 'get', requestPath),
+        summary: 'Fetch a queued request result',
+        ...(onlyActivity ? { 'x-modelschemas-activity': onlyActivity } : {}),
+        parameters: [
+          endpointParam,
+          {
+            name: 'request_id',
+            in: 'path',
+            required: true,
+            schema: { type: 'string' },
+          },
+        ],
+        responses: {
+          '200': {
+            description:
+              lean || outputsComplete
+                ? 'Queued result'
+                : `Queued result. Untyped: ${String(typed.filter((endpoint) => endpoint.output === undefined).length)} of ${String(typed.length)} models have no stored output schema, so no result union is published. GET /v1/schemas/${providerId}/{activity}/{endpoint_id}?kind=output for one model.`,
+            ...(output ? responseMedia(output, null) : {}),
+          },
+        },
+      },
+    }
+  }
+  return { paths, schemas }
+}
+
 export function operationsFor(
   providerId: string,
   endpoint: LoadedEndpoint,
@@ -493,6 +729,10 @@ export function operationsFor(
  * Assemble GET /v1/openapi/{provider}. Refuses combined documents for
  * model-grained providers; a provider-grained document is never capped by
  * path count — client generation needs the whole generation surface (#560).
+ * A model-grained queue provider (FAL) with no `?model=` is one queue
+ * document: `endpoint_id` enumerates every live model with an input schema;
+ * `?activity=` narrows the enum and inlines each model's own body schema
+ * (up to {@link MAX_QUEUE_SPEC_PATHS}).
  */
 export async function assembleProviderOpenApi(
   db: Db,
@@ -553,7 +793,31 @@ export async function assembleProviderOpenApi(
     }
   }
 
-  const loaded = await loadEndpoints(db, providerId, activity)
+  // A model-grained provider whose POST answers with a queue ack is one
+  // queue document when no model is pinned; ?activity= narrows it.
+  const queue =
+    grain === 'model' &&
+    pinnedModel === undefined &&
+    config?.connect?.siblingGet === true
+  const lean = queue && activity === undefined
+  let queueRawIds: Array<string> | undefined
+  if (queue) {
+    const listed = await loadQueueModels(db, providerId, activity)
+    const httpModels = listed.filter(
+      (model) => !hasAsyncApiFlag(model.capabilities),
+    )
+    if (listed.length > 0 && httpModels.length === 0) {
+      return {
+        ok: false,
+        status: 400,
+        code: 'spec_requires_selector',
+        message: `Activity '${activity ?? ''}' on '${providerId}' has no HTTP OpenAPI models (the listed rows publish AsyncAPI). Use GET /v1/schemas/${providerId}.`,
+      }
+    }
+    queueRawIds = httpModels.map((model) => model.rawId)
+  }
+
+  const loaded = await loadEndpoints(db, providerId, activity, !lean)
   let selected = loaded
   if (pinnedModel !== undefined) {
     const matching = loaded.filter((e) =>
@@ -568,9 +832,11 @@ export async function assembleProviderOpenApi(
         message: `Model '${pinnedModel}' is listed for '${providerId}' but has no classified generation endpoint (no path matching that raw id). List endpoints: GET /v1/schemas/${providerId}. If the model is new, wait for spec sync or POST /v1/admin/sync/${providerId}. Catalog: GET /v1/providers/${providerId}/models.`,
       }
     }
+  } else if (queueRawIds !== undefined) {
+    selected = selectQueueEndpoints(loaded, queueRawIds)
   }
 
-  if (grain === 'model' && pinnedModel === undefined) {
+  if (grain === 'model' && pinnedModel === undefined && !queue) {
     return refuseModelGrained(providerId, loaded.length)
   }
 
@@ -590,6 +856,15 @@ export async function assembleProviderOpenApi(
     }
   }
 
+  if (queue && !lean && selected.length > MAX_QUEUE_SPEC_PATHS) {
+    return {
+      ok: false,
+      status: 400,
+      code: 'spec_requires_selector',
+      message: `Activity '${activity ?? ''}' matches ${String(selected.length)} endpoints (cap ${String(MAX_QUEUE_SPEC_PATHS)}). Pin ?model={rawId}, or drop ?activity= for the whole-catalog document with untyped bodies. Counts per activity: GET /v1/providers/${providerId}.`,
+    }
+  }
+
   if (!connect || connect.servers.length === 0) {
     return {
       ok: false,
@@ -599,45 +874,67 @@ export async function assembleProviderOpenApi(
     }
   }
 
-  const modelIds = pinnedModel
-    ? [pinnedModel]
-    : (
-        await listModelsCatalog(db, {
-          provider: providerId,
-          activity,
-        })
-      ).models.map((m) => m.rawId)
+  let queueSchemas: Record<string, unknown> | undefined
   const paths: Record<string, Record<string, unknown>> = {}
-  for (const endpoint of selected) {
-    for (const op of operationsFor(
-      providerId,
-      endpoint,
-      connect,
-      modelIds,
-      pinnedModel,
-    )) {
-      const item = paths[op.path] ?? {}
-      item[op.method] = op.operation
-      paths[op.path] = item
+  if (queue) {
+    const built = buildQueueDocument(providerId, selected, connect, lean)
+    if ('empty' in built) {
+      return {
+        ok: false,
+        status: 400,
+        code: 'spec_requires_selector',
+        message: `No input schemas to publish for '${providerId}'${activity ? ` activity '${activity}'` : ''}. List endpoints: GET /v1/schemas/${providerId}.`,
+      }
+    }
+    Object.assign(paths, built.paths)
+    queueSchemas = built.schemas
+  } else {
+    const modelIds = pinnedModel
+      ? [pinnedModel]
+      : (
+          await listModelsCatalog(db, {
+            provider: providerId,
+            activity,
+          })
+        ).models.map((m) => m.rawId)
+    for (const endpoint of selected) {
+      for (const op of operationsFor(
+        providerId,
+        endpoint,
+        connect,
+        modelIds,
+        pinnedModel,
+      )) {
+        const item = paths[op.path] ?? {}
+        item[op.method] = op.operation
+        paths[op.path] = item
+      }
     }
   }
 
   const revision =
     selected.find((e) => e.specRevision)?.specRevision ?? 'current'
+  const components: Record<string, unknown> = {}
+  if (Object.keys(connect.securitySchemes).length > 0) {
+    components.securitySchemes = connect.securitySchemes
+  }
+  if (queueSchemas && Object.keys(queueSchemas).length > 0) {
+    components.schemas = queueSchemas
+  }
   const document: Record<string, unknown> = {
     openapi: '3.1.0',
     info: {
       title: `${config?.displayName ?? providerId} (modelschemas)`,
       version: revision,
-      description: `Generation-only OpenAPI assembled by modelschemas from live endpoint schemas. Call this provider directly with your own key — this is not a proxy.`,
+      description: queue
+        ? `Generation-only OpenAPI assembled by modelschemas from live endpoint schemas. Call this provider directly with your own key — this is not a proxy. One queue client: endpoint_id is the model raw id${lean ? "; bodies are untyped here, pass ?activity= to inline each model's own schema" : ", and each component schema is that model's own body (x-modelschemas-model)"}.`
+        : `Generation-only OpenAPI assembled by modelschemas from live endpoint schemas. Call this provider directly with your own key — this is not a proxy.`,
     },
     servers: connect.servers,
     ...(Object.keys(connect.securitySchemes).length > 0
-      ? {
-          security: connect.security,
-          components: { securitySchemes: connect.securitySchemes },
-        }
+      ? { security: connect.security }
       : {}),
+    ...(Object.keys(components).length > 0 ? { components } : {}),
     paths,
   }
 
diff --git a/src/server/provider-openapi.worker.test.ts b/src/server/provider-openapi.worker.test.ts
index e07fff8..8c2afbd 100644
--- a/src/server/provider-openapi.worker.test.ts
+++ b/src/server/provider-openapi.worker.test.ts
@@ -3,6 +3,8 @@ import { env } from 'cloudflare:test'
 
 import { getDb } from '#/db/index.ts'
 import type { Db } from '#/db/index.ts'
+import { eq } from 'drizzle-orm'
+
 import { endpoints, models, providers, schemaVersions } from '#/db/schema.ts'
 import { assembleProviderOpenApi } from './provider-openapi.ts'
 
@@ -58,6 +60,13 @@ beforeAll(async () => {
         method: 'POST',
         path: '/fal-ai/other',
       },
+      {
+        id: 'fal/fal-ai/retired',
+        providerId: 'fal',
+        activity: 'image',
+        method: 'POST',
+        path: '/fal-ai/retired',
+      },
     ])
     .onConflictDoNothing()
   await db
@@ -116,6 +125,26 @@ beforeAll(async () => {
         providerId: 'fal',
         rawId: 'fal-ai/flux/dev',
         activity: 'image',
+        providerMetadata: { category: 'text-to-image' },
+        firstSeenAt: NOW,
+        lastSeenAt: NOW,
+      },
+      {
+        id: 'fal-fal-ai-other',
+        providerId: 'fal',
+        rawId: 'fal-ai/other',
+        activity: 'image',
+        providerMetadata: { category: 'text-to-image' },
+        firstSeenAt: NOW,
+        lastSeenAt: NOW,
+      },
+      {
+        id: 'fal-retired',
+        providerId: 'fal',
+        rawId: 'fal-ai/retired',
+        activity: 'image',
+        providerMetadata: { category: 'text-to-image' },
+        deprecatedAt: NOW,
         firstSeenAt: NOW,
         lastSeenAt: NOW,
       },
@@ -187,14 +216,40 @@ describe('assembleProviderOpenApi', () => {
     ).toEqual(['claude-sonnet-4-5'])
   })
 
-  it('refuses a combined FAL spec and serves one model with sibling GET', async () => {
+  it('serves FAL with no selector as one untyped queue document', async () => {
     const bare = await assembleProviderOpenApi(db, 'fal')
-    expect(bare).toMatchObject({
-      ok: false,
-      status: 400,
-      code: 'spec_requires_selector',
-    })
+    expect(bare.ok).toBe(true)
+    if (!bare.ok) return
+    const barePaths = bare.document.paths as Record<
+      string,
+      {
+        post: {
+          parameters: Array<{ name: string; schema?: { enum?: Array<string> } }>
+          requestBody: {
+            content: { 'application/json': { schema: Record<string, unknown> } }
+          }
+        }
+      }
+    >
+    expect(Object.keys(barePaths).sort()).toEqual([
+      '/{endpoint_id}',
+      '/{endpoint_id}/requests/{request_id}',
+    ])
+    const bareEnum = barePaths['/{endpoint_id}']?.post.parameters.find(
+      (parameter) => parameter.name === 'endpoint_id',
+    )?.schema?.enum
+    expect(bareEnum).toContain('fal-ai/flux/dev')
+    expect(bareEnum).not.toContain('fal-ai/retired')
+    expect(
+      barePaths['/{endpoint_id}']?.post.requestBody.content['application/json']
+        .schema,
+    ).toMatchObject({ type: 'object' })
+    expect(
+      (bare.document.components as { schemas?: unknown }).schemas,
+    ).toBeUndefined()
+  })
 
+  it('serves one model with sibling GET', async () => {
     const pinned = await assembleProviderOpenApi(db, 'fal', {
       model: 'fal-ai/flux/dev',
     })
@@ -233,13 +288,401 @@ describe('assembleProviderOpenApi', () => {
     ).toBeDefined()
     expect(paths['/fal-ai/other']).toBeUndefined()
 
-    expect(
-      await assembleProviderOpenApi(db, 'fal', { activity: 'image' }),
-    ).toMatchObject({
+    const pinnedActivity = await assembleProviderOpenApi(db, 'fal', {
+      model: 'fal-ai/flux/dev',
+      activity: 'image',
+    })
+    expect(pinnedActivity.ok).toBe(true)
+    if (!pinnedActivity.ok) return
+    expect(pinnedActivity.document.paths).toHaveProperty('/fal-ai/flux/dev')
+    expect(pinnedActivity.document.paths).not.toHaveProperty('/{endpoint_id}')
+  })
+
+  it('serves one FAL queue document for an activity', async () => {
+    const empty = await assembleProviderOpenApi(db, 'fal', {
+      activity: 'chat',
+    })
+    expect(empty).toMatchObject({
       ok: false,
       status: 400,
       code: 'spec_requires_selector',
     })
+
+    const result = await assembleProviderOpenApi(db, 'fal', {
+      activity: 'image',
+    })
+    expect(result.ok).toBe(true)
+    if (!result.ok) return
+    const paths = result.document.paths as Record<
+      string,
+      {
+        post?: {
+          parameters?: Array<{
+            name: string
+            schema?: { enum?: Array<string> }
+          }>
+          requestBody?: {
+            content: {
+              'application/json': {
+                schema: { anyOf?: Array<{ $ref: string }> }
+              }
+            }
+          }
+        }
+        get?: {
+          responses?: {
+            '200'?: {
+              content?: {
+                'application/json': { schema: { $ref?: string } }
+              }
+            }
+          }
+        }
+      }
+    >
+    expect(Object.keys(paths).sort()).toEqual([
+      '/{endpoint_id}',
+      '/{endpoint_id}/requests/{request_id}',
+    ])
+    const endpointParam = paths['/{endpoint_id}']?.post?.parameters?.find(
+      (parameter) => parameter.name === 'endpoint_id',
+    )
+    expect(endpointParam).toMatchObject({ 'x-modelschemas-encode': 'raw' })
+    expect(endpointParam?.schema?.enum).toEqual([
+      'fal-ai/flux/dev',
+      'fal-ai/other',
+    ])
+    const body = paths['/{endpoint_id}']?.post?.requestBody
+    expect(body?.content['application/json'].schema.anyOf).toEqual([
+      { $ref: '#/components/schemas/input_fal-ai_flux_dev' },
+      { $ref: '#/components/schemas/input_fal-ai_other' },
+    ])
+    // fal-ai/other has no output schema, so the result stays untyped and
+    // the response says why.
+    const queued = (
+      paths['/{endpoint_id}/requests/{request_id}'] as {
+        get?: { responses?: { '200'?: { description?: string } } }
+      }
+    ).get?.responses?.['200']
+    expect(queued?.description).toContain('1 of 2 models')
+    expect(queued).not.toHaveProperty('content')
+    const schemas = (
+      result.document.components as {
+        schemas: Record<
+          string,
+          { 'x-modelschemas-model'?: string; properties?: unknown }
+        >
+      }
+    ).schemas
+    const fluxInput = schemas['input_fal-ai_flux_dev']
+    const otherInput = schemas['input_fal-ai_other']
+    expect(fluxInput?.['x-modelschemas-model']).toBe('fal-ai/flux/dev')
+    expect(fluxInput?.properties).toEqual({ prompt: { type: 'string' } })
+    expect(otherInput?.['x-modelschemas-model']).toBe('fal-ai/other')
+    // fal-ai/other has no output schema, so the result body is omitted
+    // rather than described as Flux's output.
+    expect(
+      paths['/{endpoint_id}/requests/{request_id}']?.get?.responses?.['200']
+        ?.content,
+    ).toBeUndefined()
+    expect(schemas['output_fal-ai_flux_dev']).toBeUndefined()
+    expect(schemas['input_fal-ai_retired']).toBeUndefined()
+    const info = result.document.info as { description?: string }
+    expect(info.description).toContain('x-modelschemas-model')
+  })
+
+  it('leaves a model with no generation endpoint out of the enum', async () => {
+    await db
+      .insert(models)
+      .values({
+        id: 'fal-orphan',
+        providerId: 'fal',
+        rawId: 'fal-ai/orphan',
+        activity: 'image',
+        providerMetadata: { category: 'text-to-image' },
+        firstSeenAt: NOW,
+        lastSeenAt: NOW,
+      })
+      .onConflictDoNothing()
+    const result = await assembleProviderOpenApi(db, 'fal', {
+      activity: 'image',
+    })
+    expect(result.ok).toBe(true)
+    if (!result.ok) return
+    const post = (
+      result.document.paths as Record<
+        string,
+        { post: { parameters: Array<{ schema?: { enum?: Array<string> } }> } }
+      >
+    )['/{endpoint_id}']?.post
+    expect(post?.parameters[0]?.schema?.enum).not.toContain('fal-ai/orphan')
+    await db.delete(models).where(eq(models.id, 'fal-orphan'))
+  })
+
+  it('pairs each result schema with the same endpoint id', async () => {
+    await db
+      .insert(endpoints)
+      .values([
+        {
+          id: 'fal/fal-ai/pair/a',
+          providerId: 'fal',
+          activity: 'chat',
+          method: 'POST',
+          path: '/fal-ai/pair/a',
+        },
+        {
+          id: 'fal/fal-ai/pair/b',
+          providerId: 'fal',
+          activity: 'chat',
+          method: 'POST',
+          path: '/fal-ai/pair/b',
+        },
+        {
+          id: 'fal/fal-ai/pair/no-input',
+          providerId: 'fal',
+          activity: 'chat',
+          method: 'POST',
+          path: '/fal-ai/pair/no-input',
+        },
+      ])
+      .onConflictDoNothing()
+    await db
+      .insert(models)
+      .values([
+        {
+          id: 'fal-pair-a',
+          providerId: 'fal',
+          rawId: 'fal-ai/pair/a',
+          activity: 'chat',
+          providerMetadata: { category: 'paired-image' },
+          firstSeenAt: NOW,
+          lastSeenAt: NOW,
+        },
+        {
+          id: 'fal-pair-b',
+          providerId: 'fal',
+          rawId: 'fal-ai/pair/b',
+          activity: 'chat',
+          providerMetadata: { category: 'paired-image' },
+          firstSeenAt: NOW,
+          lastSeenAt: NOW,
+        },
+        {
+          id: 'fal-pair-no-input',
+          providerId: 'fal',
+          rawId: 'fal-ai/pair/no-input',
+          activity: 'chat',
+          providerMetadata: { category: 'paired-image' },
+          firstSeenAt: NOW,
+          lastSeenAt: NOW,
+        },
+      ])
+      .onConflictDoNothing()
+    await db
+      .insert(schemaVersions)
+      .values([
+        {
+          id: 'fal/fal-ai/pair/a:input',
+          endpointId: 'fal/fal-ai/pair/a',
+          kind: 'input',
+          contentHash: `a1${'1'.repeat(62)}`,
+          schema: JSON.stringify({
+            type: 'object',
+            properties: { prompt: { const: 'a' } },
+          }),
+          createdAt: NOW,
+        },
+        {
+          id: 'fal/fal-ai/pair/a:output',
+          endpointId: 'fal/fal-ai/pair/a',
+          kind: 'output',
+          contentHash: `a2${'2'.repeat(62)}`,
+          schema: JSON.stringify({
+            type: 'object',
+            properties: { image: { const: 'a' } },
+          }),
+          createdAt: NOW,
+        },
+        {
+          id: 'fal/fal-ai/pair/b:input',
+          endpointId: 'fal/fal-ai/pair/b',
+          kind: 'input',
+          contentHash: `b1${'3'.repeat(62)}`,
+          schema: JSON.stringify({
+            type: 'object',
+            properties: { prompt: { const: 'b' } },
+          }),
+          createdAt: NOW,
+        },
+        {
+          id: 'fal/fal-ai/pair/b:output',
+          endpointId: 'fal/fal-ai/pair/b',
+          kind: 'output',
+          contentHash: `b2${'4'.repeat(62)}`,
+          schema: JSON.stringify({
+            type: 'object',
+            properties: { image: { const: 'b' } },
+          }),
+          createdAt: NOW,
+        },
+      ])
+      .onConflictDoNothing()
+
+    const result = await assembleProviderOpenApi(db, 'fal', {
+      activity: 'chat',
+    })
+    expect(result.ok).toBe(true)
+    if (!result.ok) return
+    const post = (
+      result.document.paths as Record<
+        string,
+        {
+          post?: {
+            parameters?: Array<{ schema?: { enum?: Array<string> } }>
+            requestBody?: {
+              content: {
+                'application/json': {
+                  schema: { anyOf?: Array<{ $ref: string }> }
+                }
+              }
+            }
+          }
+          get?: {
+            responses?: {
+              '200'?: {
+                content?: {
+                  'application/json': {
+                    schema: { anyOf?: Array<{ $ref: string }> }
+                  }
+                }
+              }
+            }
+          }
+        }
+      >
+    )['/{endpoint_id}']?.post
+    const get = (
+      result.document.paths as Record<
+        string,
+        {
+          get?: {
+            parameters?: Array<{ schema?: { enum?: Array<string> } }>
+            responses?: {
+              '200'?: {
+                content?: {
+                  'application/json': {
+                    schema: { anyOf?: Array<{ $ref: string }> }
+                  }
+                }
+              }
+            }
+          }
+        }
+      >
+    )['/{endpoint_id}/requests/{request_id}']?.get
+    expect(post?.parameters?.[0]?.schema?.enum).toEqual([
+      'fal-ai/pair/a',
+      'fal-ai/pair/b',
+    ])
+    expect(get?.parameters?.[0]?.schema?.enum).toEqual([
+      'fal-ai/pair/a',
+      'fal-ai/pair/b',
+    ])
+    expect(post?.requestBody?.content['application/json'].schema.anyOf).toEqual(
+      [
+        { $ref: '#/components/schemas/input_fal-ai_pair_a' },
+        { $ref: '#/components/schemas/input_fal-ai_pair_b' },
+      ],
+    )
+    expect(
+      get?.responses?.['200']?.content?.['application/json'].schema.anyOf,
+    ).toEqual([
+      { $ref: '#/components/schemas/output_fal-ai_pair_a' },
+      { $ref: '#/components/schemas/output_fal-ai_pair_b' },
+    ])
+    const schemas = (
+      result.document.components as {
+        schemas: Record<string, { 'x-modelschemas-model'?: string }>
+      }
+    ).schemas
+    expect(schemas['input_fal-ai_pair_a']?.['x-modelschemas-model']).toBe(
+      'fal-ai/pair/a',
+    )
+    expect(schemas['output_fal-ai_pair_b']?.['x-modelschemas-model']).toBe(
+      'fal-ai/pair/b',
+    )
+    expect(schemas['input_fal-ai_pair_no-input']).toBeUndefined()
+  })
+
+  it('serves an activity larger than the 40-endpoint cap as one queue document', async () => {
+    const count = 41
+    const endpointRows = Array.from({ length: count }, (_, i) => ({
+      id: `fal/bulk/m${String(i)}`,
+      providerId: 'fal',
+      activity: 'audio' as const,
+      method: 'POST',
+      path: `/bulk/m${String(i)}`,
+    }))
+    const modelRows = Array.from({ length: count }, (_, i) => ({
+      id: `fal-bulk-m${String(i)}`,
+      providerId: 'fal',
+      rawId: `bulk/m${String(i)}`,
+      activity: 'audio' as const,
+      providerMetadata: { category: 'bulk-image' },
+      firstSeenAt: NOW,
+      lastSeenAt: NOW,
+    }))
+    const schemaRows = Array.from({ length: count }, (_, i) => ({
+      id: `fal/bulk/m${String(i)}:input`,
+      endpointId: `fal/bulk/m${String(i)}`,
+      kind: 'input' as const,
+      contentHash: `${String(i).padStart(2, '0')}${'f'.repeat(62)}`,
+      schema: JSON.stringify({
+        type: 'object',
+        properties: { n: { const: i } },
+      }),
+      createdAt: NOW,
+    }))
+    for (let i = 0; i < count; i += 10) {
+      await db
+        .insert(endpoints)
+        .values(endpointRows.slice(i, i + 10))
+        .onConflictDoNothing()
+      await db
+        .insert(models)
+        .values(modelRows.slice(i, i + 10))
+        .onConflictDoNothing()
+      await db
+        .insert(schemaVersions)
+        .values(schemaRows.slice(i, i + 10))
+        .onConflictDoNothing()
+    }
+    const result = await assembleProviderOpenApi(db, 'fal', {
+      activity: 'audio',
+    })
+    expect(result.ok).toBe(true)
+    if (!result.ok) return
+    const paths = result.document.paths as Record<string, unknown>
+    expect(Object.keys(paths).sort()).toEqual([
+      '/{endpoint_id}',
+      '/{endpoint_id}/requests/{request_id}',
+    ])
+    const post = (
+      paths['/{endpoint_id}'] as {
+        post: {
+          parameters: Array<{ schema?: { enum?: Array<string> } }>
+          requestBody: {
+            content: {
+              'application/json': { schema: { anyOf: Array<unknown> } }
+            }
+          }
+        }
+      }
+    ).post
+    expect(post.parameters[0]?.schema?.enum).toHaveLength(count)
+    expect(
+      post.requestBody.content['application/json'].schema.anyOf,
+    ).toHaveLength(count)
   })
 
   it('refuses empty paths instead of returning 200', async () => {
@@ -420,6 +863,16 @@ describe('assembleProviderOpenApi', () => {
       code: 'unknown_model',
     })
     if (!result.ok) expect(result.message).toContain('AsyncAPI')
+
+    const byActivity = await assembleProviderOpenApi(db, 'fal', {
+      activity: 'video',
+    })
+    expect(byActivity).toMatchObject({
+      ok: false,
+      status: 400,
+      code: 'spec_requires_selector',
+    })
+    if (!byActivity.ok) expect(byActivity.message).toContain('AsyncAPI')
   })
 
   it('404s unknown provider and unknown model', async () => {
diff --git a/src/server/providers/connect.ts b/src/server/providers/connect.ts
index f556804..f121107 100644
--- a/src/server/providers/connect.ts
+++ b/src/server/providers/connect.ts
@@ -8,6 +8,13 @@ import type {
   SpecGrain,
 } from './types.ts'
 
+/**
+ * Queue documents narrowed with `?activity=`: FAL `video` was 653 models
+ * and `image` 638 (2026-11-04), each inlined as its own schema. The
+ * unnarrowed queue document is uncapped because it inlines none.
+ */
+export const MAX_QUEUE_SPEC_PATHS = 1024
+
 export function bearerConnect(
   serverUrl: string,
   description?: string,
diff --git a/src/server/skill.ts b/src/server/skill.ts
index 0901619..b3480ca 100644
--- a/src/server/skill.ts
+++ b/src/server/skill.ts
@@ -42,8 +42,13 @@ availability and schema questions.
      self-contained JSON Schema (URL-encode slashes in endpoint ids:
      \`chat%2Fcompletions\`). A listed model rawId also works as
      \`{endpointId}\` on grain=provider catalogs.
-   - \`GET /v1/openapi/{provider}\` — connectable generation OpenAPI
-     (FAL, currently the only model-grained provider, requires \`?model=\`)
+   - \`GET /v1/providers/{provider}\` — one provider's status, counts, and
+     live models per activity (the \`?activity=\` values).
+   - \`GET /v1/openapi/{provider}\` — connectable generation OpenAPI.
+     FAL is one queue document: the model id is the \`endpoint_id\` path
+     parameter (\`x-modelschemas-encode: raw\`, so do not percent-encode
+     slashes). Bare, the bodies are untyped; \`?activity=\` inlines each
+     model's own body schema; \`?model=\` is that one model's paths.
 3. **Authenticate for more** (5k req/h + webhooks). Easiest: the CLI —
    \`modelschemas login\` (agent-auth, autonomous mode, keys stored locally)
    or \`modelschemas login --api-key\`. Raw HTTP alternative: