> ## Documentation Index > Fetch the complete documentation index at: https://docs.reactor.inc/llms.txt > Use this file to discover all available pages before exploring further. > ## Agent Instructions > To build and serve your own model, start at /deploy/development/quickstart and /deploy/development/overview. Deploying is the default path: reactor init scaffolds a workspace, reactor auth login authenticates, and reactor model deploy registers the model, publishes the release with the weights/ folder, and activates it on Reactor's GPUs, in one command from that workspace. Docker must be running, because the publish step builds the image locally. Bump model.version in reactor.yaml before redeploying a change, because a release that already has an image is reactivated as it is. Deployment access is granted per account, so contact team@reactor.inc if a deploy is refused. Every key in reactor.yaml is documented at /deploy/platform/reactor-yaml. Model code imports reactor_runtime; Python client code imports reactor_sdk. The runtime overview explains the model interface. Running the model on your own machine with reactor run is optional and needs a GPU you attach with --gpus; /deploy/development/local-testing covers that loop and pairs a complete brightness model with a Python client test in a separate brightness-test workspace. > Reactor hosts multiple models, each with its own connect slug (modelName) and command/event schema. The video model catalog — slug, typed SDK package, and links to its schema — is at /model-api-reference/overview. Robotics policy documentation starts at /robotics/overview; The launch policies are Cosmos Nano DROID (/robotics/cosmos/nano-policy-droid/), DreamZero DROID (/robotics/dreamzero/droid/), and FLUX Action DROID (/robotics/flux-action/droid/), organized by model family and embodiment. Some models expose one slug per experience (e.g. HappyOyster); always take the slug from the model's own pages, never guess it. > Fastest path to a working app: `npx create-reactor-app my-app --model=` scaffolds a complete app with secure auth wired up. Typed TypeScript SDKs are published as @reactor-models/; Python uses the base reactor-sdk package. > Auth: exchange an API key (rk_...) for a JWT via POST https://api.reactor.inc/tokens from your server. Never put the API key in client-side code. > Append .md to any docs URL for clean Markdown. Search these docs via the MCP server at https://docs.reactor.inc/mcp. # Replace an environment > Replaces all of the environment's settings in one atomic write and creates its next revision. Use it from a client that holds the full desired state, such as a CI job. It has the rules, status codes, and response of the update, with these differences: - `image` and `instances` are required. Nothing defaults from the environment or the model. - An omitted `resources.accelerators` is `[]`, so it moves a GPU environment to CPU. - `resources` is a full object: an omitted `cpu` or `memory` value is unset. - `variables` is a full map: an omitted name is removed, and a `null` value is `400`. As on the update, an omitted `weights` keeps the current pin, and only `null` clears it. ## OpenAPI ````yaml /openapi.yaml put /projects/{proj_id}/environments/{env} openapi: 3.1.0 info: title: Reactor Platform API version: '1' description: > The Reactor Platform API mints session tokens, starts and operates sessions, negotiates media transport, and manages your account, models, and deployments. Every path is relative to `https://api.reactor.inc`. See the [Introduction](/platform-api/introduction) and [Authentication](/platform-api/authentication) pages for credentials, versioning, and errors. contact: name: Reactor support url: https://docs.reactor.inc/resources/support servers: - url: https://api.reactor.inc description: Production security: - apiKey: [] tags: - name: Status description: >- Public, unauthenticated endpoints for service health, pricing, and the values the API accepts. - name: Tokens description: > Exchange a long-lived API key for a short-lived, session-scoped token that a client can use to create and operate sessions without holding the key. - name: Users description: > Read your own user record, your account's roster, and account memberships, and manage member roles as an account admin. - name: API Keys description: >- Revoke API keys. Use the [Dashboard](https://reactor.inc/dashboard) to list or create keys. The CLI can also create a key with `reactor auth login`. - name: Sessions description: > Create, inspect, and end sessions. A session holds one running model for one client application; media and messages flow over a transport you attach to it afterwards. - name: Transport description: > Attach a client to a session. WebRTC signaling carries ICE servers, SDP, and ICE candidates; a LiveKit room connection lets the session join a room you host. - name: Files description: > Upload files a model reads, either for one running session or ahead of time for sessions you have not started yet. - name: Recordings description: > Fetch the recorded video of a session as an HLS playlist, during the session or after it ends. - name: Accounts description: > Read and manage the account your credential acts in: its name and slug, its credit balance, its effective quotas, and its accelerator capacity. - name: Invitations description: > Invite people to your account by email, and manage pending invitations. The recipient accepts from the emailed link while signed in to the dashboard. - name: Usage description: > List the sessions your account has run, filter them by status, date, model, or API key, and read exact session totals for a date window. - name: Models description: > Register, read, update, and delete the models your account serves, and list the models you can use. The `reactor model` CLI commands are built on these endpoints. - name: Releases description: > Register a model's versioned releases, push their container image, and deploy a release with a per-region instance plan. - name: Model Weights description: > Upload and inspect the weights bundle of a release: single-request uploads, multipart uploads for large files, and the server-side finalize that marks a bundle complete. - name: Model Schemas description: > Attach an OpenAPI schema to a release that describes the model's commands, messages, and tracks. Schemas of public models are readable without credentials. - name: Model Access description: | Share a private model with other accounts, one environment at a time. - name: Projects description: > A project is the deploy identity of a model you serve on Reactor. Each model is one project, with its own UUID. Creating a project also creates its `production` environment. - name: Environments description: > An environment is a named deploy target of a project, such as `production` or `staging`. Each write to an environment creates its next revision, and the newest revision is what the environment runs. Environment access grants let another account start sessions on one environment. - name: Revisions description: > A revision is an immutable, numbered snapshot of one environment: its model image, weights, resources, instances, and variables. An environment's revisions, newest first, are its change log. - name: Images description: > Model images are the container images pushed to your model's repository, named by digest. Experimental: this API can change without notice. - name: Weight Sets description: > A weight set is a version-2 weights bundle, named by its set digest. Push the files, then put the manifest. Available when weight sets are enabled for your account. Experimental: this API can change without notice. - name: Tags description: > A tag is a movable name on one model image or one weight set of a project. A release of the model is a read-only tag. Experimental: this API can change without notice. paths: /projects/{proj_id}/environments/{env}: parameters: - $ref: '#/components/parameters/ProjectProjId' - $ref: '#/components/parameters/ProjectEnv' put: tags: - Environments summary: Replace an environment description: > Replaces all of the environment's settings in one atomic write and creates its next revision. Use it from a client that holds the full desired state, such as a CI job. It has the rules, status codes, and response of the update, with these differences: - `image` and `instances` are required. Nothing defaults from the environment or the model. - An omitted `resources.accelerators` is `[]`, so it moves a GPU environment to CPU. - `resources` is a full object: an omitted `cpu` or `memory` value is unset. - `variables` is a full map: an omitted name is removed, and a `null` value is `400`. As on the update, an omitted `weights` keeps the current pin, and only `null` clears it. operationId: replaceEnvironment parameters: - $ref: '#/components/parameters/ProjectIfMatchRevision' requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/ProjectEnvironmentReplace' example: image: tag: v1.4.0 resources: accelerators: - type: NVIDIA_H100 count: 1 instances: - count: 2 variables: LOG_LEVEL: value: info API_KEY: secret: API_KEY responses: '200': description: The environment after the write. headers: ETag: description: The revision number of the returned environment in quotes. schema: type: string content: application/json: schema: $ref: '#/components/schemas/ProjectEnvironment' example: id: 4b5c6d7e-8f90-4a1b-8c2d-3e4f5a6b7c8d project_id: 0d3c6f4e-8a1b-4c2d-9e7f-1a2b3c4d5e6f name: production image: tag: v1.4.0 digest: >- sha256:9e01aa47c3b2d5f6e7a8b9c0d1e2f3a4b5c6d7e8f9a0b1c2d3e4f5a6b7c8d9e0 ref: null weights: null resources: accelerators: - type: NVIDIA_H100 count: 1 extra_args: [] node_selector: null instances: - count: 2 variables: LOG_LEVEL: value: info API_KEY: secret: API_KEY version_id: 018cc820-e222-7ba6-9f43-7cf0b7e3a112 revision: number: 5 created_by: user_id: 9a8b7c6d-5e4f-4a3b-8c2d-1e0f9a8b7c6d created_at: '2026-10-11T08:00:00Z' public: false created_at: '2026-10-09T08:53:20Z' deleted_at: null '400': $ref: '#/components/responses/ProjectEnvironmentWriteBadRequest' '401': $ref: '#/components/responses/Unauthorized' '403': $ref: '#/components/responses/ProjectEnvironmentWriteForbidden' '404': $ref: '#/components/responses/ProjectEnvironmentWriteNotFound' '409': $ref: '#/components/responses/ProjectEnvironmentWriteConflict' '412': $ref: '#/components/responses/ProjectEtagMismatch' '429': $ref: '#/components/responses/ProjectEnvironmentWriteQuota' '503': $ref: '#/components/responses/ProjectEnvironmentWriteUnavailable' components: parameters: ProjectProjId: name: proj_id in: path required: true description: The project's ID, a UUID. It is not the model's ID. schema: type: string format: uuid ProjectEnv: name: env in: path required: true description: >- The environment's ID (a UUID) or its name, such as `production`. A name matches only a live environment; a deleted environment is reachable by its ID only. schema: type: string ProjectIfMatchRevision: name: If-Match in: header required: false description: >- The `ETag` of your last read, a revision number in quotes. If the environment's newest revision differs, the write is `412 etag_mismatch` and nothing is written. A weak tag, a list, or `*` is `400`. schema: type: string example: '"3"' schemas: ProjectEnvironmentReplace: type: object required: - image - instances additionalProperties: false description: >- The full desired settings. Read-only fields are accepted only with their current values. properties: image: $ref: '#/components/schemas/ProjectImageInput' weights: type: - string - 'null' description: As on the update. Omitted keeps the current pin; `null` clears it. resources: allOf: - $ref: '#/components/schemas/ProjectResources' description: >- A full object. An omitted `accelerators` is `[]` (CPU); an omitted `cpu` or `memory` value is unset. extra_args: type: array items: type: string instances: type: array items: $ref: '#/components/schemas/ProjectInstance' variables: allOf: - $ref: '#/components/schemas/ProjectVariables' description: The full map. An omitted name is removed. deployment_strategy: type: string enum: - ROLLING - REPLACE_ALL default: ROLLING interrupt_sessions: type: boolean default: false ProjectEnvironment: type: object required: - id - project_id - name - image - weights - resources - extra_args - node_selector - instances - revision - public - created_at - deleted_at properties: id: type: string format: uuid description: >- The environment's ID. It is neither the project's nor the model's ID. project_id: type: string format: uuid name: type: string image: oneOf: - $ref: '#/components/schemas/ProjectImageRef' - type: 'null' description: >- The current image. For `production`, the deployed release's image. Null for another environment before its first revision. weights: oneOf: - $ref: '#/components/schemas/ProjectWeightsRef' - type: 'null' description: The pinned weights. Null when none are pinned. resources: $ref: '#/components/schemas/ProjectResources' extra_args: type: array items: type: string description: Extra arguments passed to the model container. node_selector: type: - string - 'null' description: Read-only. instances: type: array items: $ref: '#/components/schemas/ProjectInstance' variables: $ref: '#/components/schemas/ProjectVariables' revision: $ref: '#/components/schemas/ProjectRevisionRef' public: type: boolean description: >- When true, any authenticated caller can open sessions on `production`. It grants no read. Read-only for customers. created_at: type: string format: date-time deleted_at: type: - string - 'null' format: date-time description: >- When the environment was deleted. A deleted environment takes no writes. ProjectImageInput: type: object description: > The image to run. Send `tag`, `ref`, both (they must name the same image), or `digest` alone. A `tag` is a model image tag, or a release of the model; a tag wins over a release of the same name. A 12 to 64 character lowercase hex `tag` that no tag has names the image whose digest it starts. A `digest` is `sha256:<64 hex>` or a 12+ character hex prefix. A `ref`, a `digest`, or such a hex `tag` runs the image with no release. properties: tag: type: string example: v1.3.0 ref: type: string description: A digest reference, `…@sha256:<64 hex>`, to an image of the project. digest: type: string example: >- sha256:d138642b1cd5ea5bba6ed5655fa1d7dfced33889d149b094b2c32c32f09ed3da ProjectResources: type: object description: What each instance gets. properties: accelerators: type: - array - 'null' maxItems: 4 items: $ref: '#/components/schemas/ProjectAccelerator' description: > The accelerator configurations, in preference order, with no repeats. `[]` is CPU. A list longer than one is not allowed on `production`. Null on a GPU model that has no stored GPU type, until a write sets it. cpu: $ref: '#/components/schemas/ProjectQuantities' memory: $ref: '#/components/schemas/ProjectQuantities' ProjectInstance: type: object required: - count description: > One entry of the instances declaration, in one of two forms. **Per region**: each entry names `count` instances in the one region of `regions.only`, totalling at most 1000. **Total**: exactly one entry with the total `count` (at most 1000), placed by Reactor; `regions.prefer` names regions to fill first, `regions.only` restricts placement. A list replaces the current one: an omitted region scales to zero, and `[]` scales everything to zero. properties: count: type: integer minimum: 0 regions: type: object properties: only: type: array items: type: string description: The only regions to use. prefer: type: array items: type: string description: >- Regions to fill first, in order. Total form only, and not with `only`. ProjectVariables: type: object description: > Variables keyed by name. Names use letters, digits, `_`, `-`, and `.`, with no leading digit. Reserved names such as `MODEL_NAME` and any `REACTOR_` name are refused. additionalProperties: $ref: '#/components/schemas/ProjectVariable' ProjectImageRef: type: object description: The model image an environment or revision runs. properties: tag: type: - string - 'null' description: >- The release the revision carries. Null when it runs an image with no release. digest: type: - string - 'null' description: The image digest, `sha256:<64 hex>`. ref: type: - string - 'null' description: Always null for customer callers. ProjectWeightsRef: type: object description: The weights an environment or revision pins. properties: tag: type: - string - 'null' description: >- The release that named the weights. Null when a write named a digest or a tag. ref: type: - string - 'null' description: A pinned `hf://owner/repo@` reference, or null. digest: type: - string - 'null' description: >- The set digest, `sha256:<64 hex>`. Null for a version-1 bundle or a bare `hf://` reference. ProjectRevisionRef: type: - object - 'null' description: The current revision. Null before the first revision. properties: number: type: integer created_by: $ref: '#/components/schemas/ProjectCreatedBy' created_at: type: string format: date-time Error: type: object required: - error properties: error: type: string description: >- A stable error code (for example `not_found`) on newer routes, or a sentence on older ones. When `message` is present, branch on `error`. message: type: string description: >- A human-readable explanation. Do not parse it; its wording may change. resource: type: string description: The type of the resource a `not_found` refers to. reason: type: string description: >- A more specific cause for `failed_precondition`, `aborted`, and some `invalid_request` errors. ProjectIncompleteError: allOf: - $ref: '#/components/schemas/Error' - type: object properties: missing_fields: type: array items: type: string description: >- For `environment_incomplete`, each missing setting by its path, such as `image` or `resources.accelerators`. ProjectQuotaError: allOf: - $ref: '#/components/schemas/Error' - type: object properties: quota_type: type: string description: The quota that was hit, such as `registry_reads`. limit: type: string description: The limit, as written. current: type: string description: The requested or current value, as written. field: type: string description: >- The request member over the limit, such as `resources.cpu.limit`. ProjectAccelerator: type: object required: - type - count properties: type: type: string description: A GPU type, as listed by `GET /gpu_types`. example: NVIDIA_H100 count: type: integer minimum: 1 description: GPUs per instance. A time-sliced type takes 1 only. ProjectQuantities: type: object properties: request: type: - string - 'null' description: >- A Kubernetes quantity, such as `2` or `16Gi`. `null` clears it on a write. limit: type: - string - 'null' description: A Kubernetes quantity. A request above its limit is `400`. ProjectVariable: type: object description: > One environment variable: `{"value": "..."}` for a plain value, or `{"secret": ""}` for a secret binding. A read adds the pinned `version_id`. A secret value never appears here. properties: value: type: string description: A plain value. secret: type: string description: The name of a secret to bind. A write pins its latest version. version_id: type: string format: uuid description: The pinned secret version. Send it back unchanged to keep the pin. ProjectCreatedBy: type: - object - 'null' description: >- The user who made the write. An API-key write records the key's owner. Null for a system write. properties: user_id: type: string format: uuid responses: ProjectEnvironmentWriteBadRequest: description: > `invalid_request`: a malformed body, an unknown or read-only field with a new value, a rule of a setting broken, or an `If-Match` that is not one quoted revision number. The message names the field. With `reason` `invalid_id`, `proj_id` is not a UUID. `invalid_deployment_plan`: a region of a per-region `instances` entry is unknown. content: application/json: schema: $ref: '#/components/schemas/Error' Unauthorized: description: The credential is missing, invalid, or expired. content: application/json: schema: $ref: '#/components/schemas/Error' ProjectEnvironmentWriteForbidden: description: > The caller lacks the `account_admin` or `developer` role, Serve access is not enabled for the account, the project does not exist or belongs to another account, or the body changes a setting reserved to Reactor (`node_selector`, `public`). content: application/json: schema: $ref: '#/components/schemas/Error' ProjectEnvironmentWriteNotFound: description: > `not_found`. `resource` names what is missing: `environment`; `image` or `release` for an `image` that names nothing in the project; `weights` for a `weights` value that names no weight set; `secret` for a binding to a secret that does not exist. content: application/json: schema: $ref: '#/components/schemas/Error' ProjectEnvironmentWriteConflict: description: > Nothing was written. `failed_precondition`, with one of these `reason` values: - `deleted`: the environment is deleted. - `environment_incomplete`: the result has no `image`, `instances`, or `resources.accelerators`. `missing_fields` lists them. - `tag_has_no_image`: the release has no published image. - `release_not_semver`: the release's tag is not a semver version. - `release_required`: this model's `production` needs a release name in `image.tag`. - `weights_pin_unavailable`: this model's `production` cannot pin a weight set yet. Name a release that holds the set. - `weights_need_release`: the revision would keep weights with no set digest. Send `weights`, or name a release. - `set_objects_off`: `weights` needs weight sets, which are not enabled for your account. - `plain_variables_too_large`: plain variable values exceed 32 KiB. - `resolved_environment_too_large`: plain and secret values together exceed 256 KiB. - `deployment_plan_unplaceable`: a region cannot hold its instances on the listed accelerators. Also `accelerator_not_supported_in_region` (no cluster in a planned region runs a listed type), `accelerator_quota_exceeded` (the plan exceeds the account's quota in a region), `weights_bundle_unverifiable`, and `aborted` with `reason` `concurrent_update` (another write landed first; retry). content: application/json: schema: $ref: '#/components/schemas/ProjectIncompleteError' example: error: failed_precondition reason: environment_incomplete message: 'environment staging is not complete: set image' missing_fields: - image ProjectEtagMismatch: description: >- `etag_mismatch`: `If-Match` names a revision that is not the newest one. Nothing was written. Read the environment again, then decide. content: application/json: schema: $ref: '#/components/schemas/Error' example: error: etag_mismatch message: environment staging is at revision 4, not 3 ProjectEnvironmentWriteQuota: description: > `quota_exceeded`. Either the revision's CPU or memory is above the account's limit (`field` names the request member; no `Retry-After`), or `image` names an unrecorded digest and the account is over its registry read limit (retry after `Retry-After`). headers: Retry-After: schema: type: integer description: Seconds to wait for the registry read limit only. content: application/json: schema: $ref: '#/components/schemas/ProjectQuotaError' ProjectEnvironmentWriteUnavailable: description: > `unavailable` or `accelerator_inventory_unavailable`: a dependency could not be read, or live sessions could not be ended for `interrupt_sessions`. Nothing was written; retry. content: application/json: schema: $ref: '#/components/schemas/Error' securitySchemes: apiKey: type: http scheme: bearer description: >- A Reactor API key (`rk_...`), sent as `Authorization: Bearer rk_...`. Keep it on a server. ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.