> ## Documentation Index > Fetch the complete documentation index at: https://docs.agentsfleet.net/llms.txt > Use this file to discover all available pages before exploring further. # List available models > Returns models available to the tenant. Each item includes its provider, context limit, and usage fields. An empty library returns an empty array. A bounded keyset page ordered by normalized `model_id`, then normalized `provider`, then an opaque row identity. Every comparison runs under `COLLATE "C"`, so the order is byte-wise and stable across locales. Follow `next_cursor` to read the whole catalogue; a client that reads only the first page silently loses every model past it. The response is conditionally revalidated. Every answer carries an `ETag` over the exact bytes served, plus `Cache-Control: private, no-cache` and `Vary: Authorization`. Send the tag back as `If-None-Match` and a match answers `304` with no body and the same headers. `private` is what stops a shared proxy handing one tenant's response to another even though the payload is identical for every authorized caller. ## OpenAPI ````yaml https://raw.githubusercontent.com/agentsfleet/agentsfleet/main/public/openapi.json get /v1/models openapi: 3.1.0 info: title: agentsfleet Control Plane API version: 1.0.0 description: API for managing workspaces, fleets, triggers, and runs. contact: name: agentsfleet url: https://agentsfleet.net servers: - url: https://api.agentsfleet.net description: Production security: - BearerAuth: [] tags: - name: Health description: System status and readiness probes x-mintlify: navigation: Health - name: Authentication description: Command Line Interface (CLI) sign-in sessions and GitHub sign-in x-mintlify: navigation: Authentication - name: Workspaces description: Create and manage workspaces x-mintlify: navigation: Workspaces - name: Tenant description: Manage tenant provider settings and models x-mintlify: navigation: Tenant - name: auth-identity-events description: >- Process signed account events from Clerk. This route is not for customer fleets. x-mintlify: navigation: Auth Identity Events - name: Fleets description: Create, configure, and run fleets x-mintlify: navigation: Fleets - name: Schedules description: Manage and receive scheduled Fleet events x-mintlify: navigation: Schedules - name: Memory description: Read and search saved fleet memory x-mintlify: navigation: Memory - name: Secrets description: Manage named secrets for fleets in a workspace x-mintlify: navigation: Secrets - name: Admin description: Manage shared provider keys and models x-mintlify: navigation: Admin - name: Billing description: Read tenant balances and charge records x-mintlify: navigation: Billing - name: Integration Grants description: Manage fleet access to third-party services x-mintlify: navigation: Integration Grants - name: Connectors description: Connect workspaces to third-party providers x-mintlify: navigation: Connectors - name: Fleet Keys description: Manage keys that let external clients call a fleet x-mintlify: navigation: Fleet Keys - name: API Keys description: Manage tenant admin API keys x-mintlify: navigation: API Keys - name: Webhooks description: Receive signed events and approval decisions x-mintlify: navigation: Webhooks - name: Approvals description: List, inspect, and resolve pending approvals x-mintlify: navigation: Approvals - name: Fleet description: Enroll and manage runners x-mintlify: navigation: Fleet - name: Fleet Bundles description: Manage fleet source bundles in a workspace x-mintlify: navigation: Fleet Bundles - name: Fleet library description: Browse and add reusable fleet sources x-mintlify: navigation: Fleet library - name: Model Library description: List models available to the tenant x-mintlify: navigation: Model Library paths: /v1/models: get: tags: - Model Library summary: List available models description: > Returns models available to the tenant. Each item includes its provider, context limit, and usage fields. An empty library returns an empty array. A bounded keyset page ordered by normalized `model_id`, then normalized `provider`, then an opaque row identity. Every comparison runs under `COLLATE "C"`, so the order is byte-wise and stable across locales. Follow `next_cursor` to read the whole catalogue; a client that reads only the first page silently loses every model past it. The response is conditionally revalidated. Every answer carries an `ETag` over the exact bytes served, plus `Cache-Control: private, no-cache` and `Vary: Authorization`. Send the tag back as `If-None-Match` and a match answers `304` with no body and the same headers. `private` is what stops a shared proxy handing one tenant's response to another even though the payload is identical for every authorized caller. operationId: get_model_library parameters: - name: limit in: query required: false description: Rows per page, 1..100. Defaults to 50. schema: type: integer minimum: 1 maximum: 100 default: 50 - name: starting_after in: query required: false description: >- Opaque cursor from a previous page's `next_cursor`. Bound to the filters and page size that produced it. A cursor this endpoint never issued is `UZ-LIBRARY-001`. A real cursor carrying different filters or a different page size is `UZ-LIBRARY-002`, rather than a silently different page. Only the sort boundary is read from it — the filters applied are always the request's. schema: type: string - name: provider in: query required: false description: >- Exact match on the normalized provider. At most 128 bytes once normalized; empty after normalization is treated as absent. schema: type: string maxLength: 128 responses: '200': description: One page of the model library catalogue. headers: ETag: description: Strong validator over the exact bytes served. schema: type: string Cache-Control: description: Always `private, no-cache` — store, but revalidate before reuse. schema: type: string Vary: description: Always `Authorization`. schema: type: string content: application/json: schema: type: object required: - version - models - total - next_cursor properties: version: type: string description: >- Date of the most recently updated model, formatted as YYYY-MM-DD in UTC. models: type: array description: >- The page's rows. Named `models` rather than `items` because renaming a shipped `/v1` field is a breaking change. `total` and `next_cursor` are added beside it, so the page is navigable without breaking an existing client. items: type: object required: - id - provider - context_cap_tokens - input_nanos_per_mtok - cached_input_nanos_per_mtok - output_nanos_per_mtok properties: id: type: string description: >- The model_id. Provider hosting is encoded in the id itself. provider: type: string context_cap_tokens: type: integer format: int32 input_nanos_per_mtok: type: integer format: int64 cached_input_nanos_per_mtok: type: integer format: int64 output_nanos_per_mtok: type: integer format: int64 total: type: integer nullable: true description: >- Always null. Counting a keyset page costs the scan this pagination exists to avoid. The key is present rather than omitted, so no client need special-case its absence. next_cursor: type: string nullable: true description: Cursor for the next page, or null on the last. '304': description: >- The caller's `If-None-Match` matched the current tag. No body. Carries the same `ETag`, `Cache-Control` and `Vary` as the `200`. A `304` that omitted them would tell a cache to stop revalidating the very response it is holding. headers: ETag: description: Strong validator over the bytes the caller already holds. schema: type: string Cache-Control: description: Always `private, no-cache`. schema: type: string Vary: description: Always `Authorization`. schema: type: string '400': $ref: '#/components/responses/Error' '401': $ref: '#/components/responses/Error' '503': $ref: '#/components/responses/Error' default: $ref: '#/components/responses/Error' components: responses: Error: description: RFC 7807 problem detail error response content: application/problem+json: schema: $ref: '#/components/schemas/ErrorBody' schemas: ErrorBody: type: object description: RFC 7807 problem detail error response (application/problem+json) required: - docs_uri - title - detail - error_code - request_id properties: docs_uri: type: string format: uri description: Stable link to documentation for this error code example: https://docs.agentsfleet.net/api-reference/error-codes#UZ-AGT-009 title: type: string description: Short human-readable label, same for every occurrence of this code example: Fleet not found detail: type: string description: Instance-specific context describing what went wrong example: No fleet with id 'abc123' in this workspace. error_code: type: string description: Machine-readable error code example: UZ-AGT-009 request_id: type: string description: Correlation ID for this request current_state: type: string description: >- Resource state that blocked the requested change. Present only on 409 responses. example: paused user_message: type: string description: > Plain message safe to show to a user. Present only when an error defines one. Use `detail` or `title` when this field is absent. example: >- We couldn't find that Fleet. It may have been deleted, or the ID doesn't match one in this workspace. etag: type: string description: >- Resource version returned on 412 responses so the caller can refetch and retry. securitySchemes: BearerAuth: type: http scheme: bearer bearerFormat: JWT description: >- Obtain a token via the CLI auth flow (POST /v1/auth/sessions) or GitHub OAuth ````