> ## 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 > Reactor hosts multiple models, each with its own connect slug (modelName) and command/event schema. The catalog of every model — slug, typed SDK package, and links to its schema — is at /model-api-reference/overview. 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. # reactor model publish > Build, push, and prepare a model release for deployment ```sh theme={"theme":{"light":"github-light","dark":"github-dark-high-contrast"}} reactor model publish [:] [flags] ``` ## Description Publish a release of a model - image + weights, all in one command: 1. registers the release tag (idempotent), 2. mints short-lived ECR push credentials, 3. builds the image (or loads `--source`) and pushes to the partner slot, 4. records the digested image ref on the server, 5. optionally uploads weights (see `--weights` below). **Workspace-aware:** When run from a workspace (any directory under one containing a `reactor.yaml`, stopping at the enclosing repository root) with no positional argument, the model name and release tag are taken from the file's `model.name` and `model.version` fields, so you can simply type `reactor model publish`. An explicit `:` on the command line can override the workspace release tag, but its `` part must match the workspace `model.name` or the command is rejected. **Image source (step 3):** The image is built from reactor.yaml's `build:` block, with the Dockerfile generated in memory and never written to the workspace. A Dockerfile on disk takes the build over: the workspace ./Dockerfile is built verbatim when it exists, `-f`/`--dockerfile` builds an explicit path, and either way reactor.yaml's build fields are ignored with a stderr note naming them. `--no-dockerfile` builds from reactor.yaml even when a ./Dockerfile exists, and is mutually exclusive with `-f`. `--source` skips the build entirely and loads an existing image. `--build-secret` supplies a build secret (id=\,env=\ grammar - see `reactor build --help`); required whenever the Dockerfile (on-disk or generated) declares a secret mount, such as a build.run hook with one. Ignored with `--source`. **Weights (step 5):** * `--weights `: Upload weights from this file or directory for this release. Without it, the weights step is skipped and `reactor model deploy` reuses the previously deployed release's weights (upload new ones later with `reactor weights upload` if they changed). * `--force`: Re-upload weights even when the target release already has a matching bundle on the server (only meaningful with `--weights`). **Examples:** ```sh theme={"theme":{"light":"github-light","dark":"github-dark-high-contrast"}} reactor model publish # workspace: read name+version from reactor.yaml reactor model publish my-model:v1.0.0-gabc1234 reactor model publish my-model:v1.0.0 --source my-org/my-image:v1.0.0 reactor model publish my-model:v1.0.0 -f infra/Dockerfile.runtime reactor model publish my-model:v1.0.0 --no-dockerfile # ignore the on-disk ./Dockerfile reactor model publish my-model:v1.0.0 --weights ./checkpoints/ reactor model publish my-model:v1.0.0 --weights ./checkpoints/ --force reactor model publish my-model:v1.0.0 --dry-run ``` ## Options | Flag | Description | | ---------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `--build-secret stringArray` | Build secret, id=\,env=\\|src=\ (repeatable; e.g. id=github\_token,env=GITHUB\_TOKEN). Ignored with `--source`. | | `-f, --dockerfile string` | Path to an explicit Dockerfile (relative to the build context = cwd); defaults to the workspace ./Dockerfile when one exists | | `--dry-run` | Print the resolved plan and exit; no side effects | | `--force` | Re-upload weights even when the target release already has a matching bundle on the server | | `--no-dockerfile` | Build from reactor.yaml, ignoring an on-disk Dockerfile | | `--no-register` | Fail if the model is not already registered, instead of auto-creating it from the workspace reactor.yaml (publish registers on miss by default) | | `--platform string` | Target build platform (default linux/amd64). Overrides build.platform in reactor.yaml; an explicit 'FROM `--platform`=' in the Dockerfile still wins for that stage. Ignored with `--source`. | | `--source string` | Use an existing image reference instead of building from a Dockerfile | | `--weights string` | Path to a weights file or directory to upload for this release (omit to reuse the previously deployed release's weights at deploy time) | | `--workers int` | Parallel workers for bundle hashing and per-file uploads during the weights step (default 4) | ## Global options | Flag | Description | | ----------------- | ------------------------------------------------------------------------------------- | | `--server string` | Reactor API server URL (default "[https://api.reactor.inc](https://api.reactor.inc)") | | `-v, --verbose` | Enable verbose output | ## See also * [reactor model](/deploy/cli-reference/reactor_model) - Model management commands