@@ -181,8 +181,8 @@ Relevant paths: templates/start, run/task, dependencies/tools, plugin/self.
 
 Relevant paths: workload/artifact commands, manifest/wapi/wizard/sync packages.
 
-- Set `DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA=true` for config/up help or local wizard
-  tests that need it. Check help and required flag/ID/spec validation.
+- Check help and required flag/ID/spec validation; config, up and promote need
+  no feature gate.
 - Use scratch Dockerfiles, `.datarobot.yaml`, `.env`, and project state built from
   current documented formats. Verify local schema rejection, file preservation,
   and clear remediation without making DataRobot requests.
 
@@ -66,8 +66,8 @@ feature_flags:
   provider: custom-environment
   how_to_override: >
     Set DATAROBOT_CLI_FEATURE_<NAME>=true for the selected flow only.
-    Workload config/up use DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA; enclave uses
-    DATAROBOT_CLI_FEATURE_ENCLAVE. A CLI gate does not grant platform entitlements.
+    Enclave uses DATAROBOT_CLI_FEATURE_ENCLAVE; workload commands are ungated.
+    A CLI gate does not grant platform entitlements.
 
 integrations:
   datarobot:
 
@@ -7,7 +7,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 ## Changed
 
-- The feature gate for `dr workload config`, `up` and `promote` is now `DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA=true`; the three commands are alpha and stay gated. `DATAROBOT_CLI_FEATURE_WORKLOAD` is no longer read, so a script that set it must set the new variable to keep seeing them.
+- `dr workload config`, `up` and `promote` are released and no longer behind a feature gate: they are in `--help`, completion and dispatch with nothing set. A script that still sets `DATAROBOT_CLI_FEATURE_WORKLOAD` or `DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA` keeps working; neither is needed. `DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA` stays reserved for workload features still in alpha.
 
 ## Added
 
 
@@ -95,9 +95,7 @@ wl::init_env() {
     # configured endpoint wins (env binding would otherwise shadow config).
     [[ -n "${DATAROBOT_ENDPOINT:-}" ]] && export DATAROBOT_ENDPOINT
 
-    # The workload feature gate is CLI-side and never in drconfig.yaml, so the
-    # suite always sets it. Non-interactive keeps prompts out of CI/local runs.
-    export DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA=true
+    # Non-interactive keeps prompts out of CI/local runs.
     export DATAROBOT_CLI_NON_INTERACTIVE=1
 
     # Per-run scratch directory; removed on exit (see wl::register_cleanup).
 
@@ -423,21 +423,17 @@ func TestArtifactCommandPresentByDefault(t *testing.T) {
 		"artifact command should be present by default now that it is no longer feature-gated")
 }
 
-// TestWorkloadUpAndConfigAbsentByDefault verifies that the two workload
-// subcommands still behind DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA are missing from the
-// tree while it is unset: not hidden, absent, so they are out of help,
-// completion and dispatch alike. The env var is neutralized for the same
-// reason as above.
-func TestWorkloadUpAndConfigAbsentByDefault(t *testing.T) {
+// TestWorkloadUpConfigPromotePresentByDefault verifies that config, up and
+// promote are released: present in the tree with the alpha gate unset. The
+// env var is neutralized for the same reason as above.
+func TestWorkloadUpConfigPromotePresentByDefault(t *testing.T) {
 	t.Setenv("DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA", "")
 
 	root := newIsolatedRootCmd()
 
-	require.NotNil(t, findCommandByPath(root.Command, "dr workload"))
-
-	for _, path := range []string{"dr workload up", "dr workload config"} {
-		assert.Nil(t, findCommandByPath(root.Command, path),
-			"%s should be absent when DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA is unset", path)
+	for _, path := range []string{"dr workload up", "dr workload config", "dr workload promote"} {
+		assert.NotNil(t, findCommandByPath(root.Command, path),
+			"%s should be present without DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA", path)
 	}
 }
 
 
@@ -19,7 +19,6 @@ import (
 	"testing"
 
 	"github.com/datarobot/cli/cmd/enclave"
-	"github.com/datarobot/cli/cmd/workload"
 	"github.com/spf13/cobra"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
@@ -49,8 +48,7 @@ var expectedTrackedCommands = []string{
 	"dr plugin update",
 
 	// Pipelines, workloads and artifacts are GA, so no feature gate hides them
-	// from the live RootCmd and they need no separate standalone list. The two
-	// workload leaves still behind a gate are checked further down.
+	// from the live RootCmd and they need no separate standalone list.
 	"dr pipeline create",
 	"dr pipeline get",
 	"dr pipeline clone",
@@ -99,6 +97,9 @@ var expectedTrackedCommands = []string{
 	"dr workload status",
 	"dr workload endpoint",
 	"dr workload logs",
+	"dr workload config",
+	"dr workload up",
+	"dr workload promote",
 
 	"dr artifact create",
 	"dr artifact get",
@@ -190,39 +191,6 @@ func leafCommands(root *cobra.Command) []*cobra.Command {
 	return leaves
 }
 
-// expectedGatedWorkloadTrackedCommands enumerates the `dr workload` leaves
-// still behind DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA. cli.CommandAdder leaves them
-// out of the tree while the variable is unset (the default in CI), so the
-// test below sets it and walks a freshly-built subtree from workload.Cmd()
-// rather than the global RootCmd.
-//
-// Paths are relative to workload.Cmd() (no "dr" prefix) because
-// findCommandByPath matches against the root's Name(), which is "workload"
-// for the standalone subtree.
-var expectedGatedWorkloadTrackedCommands = []string{
-	"workload config",
-	"workload up",
-}
-
-// TestTelemetryWiring_GatedWorkloadCommandsTracked enables the workload gate,
-// builds the subtree and asserts each gated leaf has the "telemetry"
-// annotation set by telemetry.Track / TrackWith.
-func TestTelemetryWiring_GatedWorkloadCommandsTracked(t *testing.T) {
-	t.Setenv("DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA", "true")
-
-	workloadRoot := workload.Cmd()
-
-	for _, path := range expectedGatedWorkloadTrackedCommands {
-		t.Run("dr "+path, func(t *testing.T) {
-			cmd := findCommandByPath(workloadRoot, path)
-			require.NotNilf(t, cmd, "command %q not found in workload subtree", path)
-
-			assert.Containsf(t, cmd.Annotations, "telemetry",
-				"command %q must be wired to telemetry via telemetry.Track / TrackWith", path)
-		})
-	}
-}
-
 // expectedEnclaveTrackedCommands enumerates leaf commands under `dr enclave`
 // that must be wired to fire a telemetry event. Like the workload subtree,
 // `dr enclave` is hidden from the live RootCmd by cli.CommandAdder when
 
@@ -31,7 +31,6 @@ import (
 	"github.com/datarobot/cli/cmd/workload/stop"
 	"github.com/datarobot/cli/cmd/workload/up"
 	"github.com/datarobot/cli/internal/cli"
-	"github.com/datarobot/cli/internal/features"
 	"github.com/spf13/cobra"
 )
 
@@ -46,22 +45,16 @@ func Cmd() *cobra.Command {
 Manage and monitor workloads in your deployment infrastructure.`,
 	}
 
-	// `config`, `up` and `promote` are alpha and stay behind
-	// DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA=true; the adder leaves them out at
-	// registration while it is unset, so they are absent from help, completion
-	// and dispatch rather than merely hidden.
-	gated := func(c *cobra.Command) *cobra.Command {
-		features.SetGate(c, "workload-alpha")
-
-		return c
-	}
-
+	// Registered through the adder so a future alpha subcommand can be gated
+	// with features.SetGate(c, "workload-alpha"), which keeps it out of help,
+	// completion and dispatch until DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA=true.
+	// Nothing here is gated today.
 	adder := &cli.CommandAdder{Command: cmd}
 	adder.AddCommand(
 		// Setup, and the one subcommand here that never calls the API: it
 		// writes the committed .datarobot.yaml that `up` deploys from. Listed
 		// apart from the verbs below so it does not read as one of them.
-		gated(config.Cmd()),
+		config.Cmd(),
 
 		// The workload itself is the primary resource: direct verbs, like
 		// `dr pipeline create|get|...`.
@@ -73,12 +66,12 @@ Manage and monitor workloads in your deployment infrastructure.`,
 		get.Cmd(),
 		list.Cmd(),
 		logs.Cmd(),
-		gated(promote.Cmd()),
+		promote.Cmd(),
 		settings.Cmd(),
 		start.Cmd(),
 		status.Cmd(),
 		stop.Cmd(),
-		gated(up.Cmd()),
+		up.Cmd(),
 	)
 
 	return cmd
 
@@ -32,48 +32,27 @@ func TestCmd_NotFeatureGated(t *testing.T) {
 
 	// The workload root is generally available: it must carry no feature-gate
 	// annotation, or cli.CommandAdder drops it from the root command tree
-	// unless DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA is set. The gate now lives on two
-	// of its subcommands instead; see TestCmd_GatesUpAndConfig below and
-	// TestWorkloadCommandPresentByDefault in cmd/root_test.go.
+	// unless DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA is set.
 	assert.NotContains(t, cmd.Annotations, features.AnnotationKey,
 		"workload is GA and must not carry a %q annotation", features.AnnotationKey)
 }
 
-// TestCmd_GatesUpAndConfig pins which subcommands the workload gate still
-// covers: `up` and `config` are registered only when
-// DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA is set, every other verb always. The gate is
-// applied at registration inside Cmd(), so building the subtree is enough.
-func TestCmd_GatesUpAndConfig(t *testing.T) {
-	always := []string{"create", "delete", "endpoint", "get", "list", "logs", "start", "status", "stop"}
-	gated := []string{"config", "promote", "up"}
+// Every verb shipped so far is released: with the alpha gate unset, all of
+// them are registered, config, up and promote included. Scripts that never
+// set the gate, or set the name it had before, must keep finding them.
+func TestCmd_RegistersEveryVerbWithoutTheGate(t *testing.T) {
+	t.Setenv("DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA", "")
 
-	tests := []struct {
-		name      string
-		gate      string
-		wantGated bool
-	}{
-		{name: "gate unset leaves up and config out", gate: "", wantGated: false},
-		{name: "gate set registers up and config", gate: "true", wantGated: true},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			t.Setenv("DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA", tt.gate)
-
-			registered := map[string]bool{}
+	registered := map[string]bool{}
 
-			for _, sub := range Cmd().Commands() {
-				registered[sub.Name()] = true
-			}
-
-			for _, name := range always {
-				assert.True(t, registered[name], "dr workload %s must always be registered", name)
-			}
+	for _, sub := range Cmd().Commands() {
+		registered[sub.Name()] = true
+	}
 
-			for _, name := range gated {
-				assert.Equal(t, tt.wantGated, registered[name],
-					"dr workload %s registered with DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA=%q", name, tt.gate)
-			}
-		})
+	for _, name := range []string{
+		"config", "create", "delete", "diagnose", "endpoint", "events", "get", "list",
+		"logs", "promote", "settings", "start", "status", "stop", "up",
+	} {
+		assert.True(t, registered[name], "dr workload %s must be registered without the gate", name)
 	}
 }
@@ -2,9 +2,6 @@
 
 The file `dr workload config` writes and `dr workload up` deploys from. This page is the reference for the file itself and for what `up` does with it. The commands and their flags are documented in [`dr workload`](workload.md).
 
-> [!NOTE]
-> `config` and `up` are alpha and behind a feature gate. Set `DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA=true` to use them.
-
 ## What the file is
 
 `.datarobot.yaml` sits at the root of your project and is the **workload-create spec, verbatim**: the same document `dr workload create --spec-file` accepts, with every field documented in the [spec reference](workload-spec.md#workload-spec). Keys the CLI does not know are passed to the platform untouched, so a platform addition needs no CLI release.
 
@@ -212,8 +212,6 @@ A workload whose id is specified in the manifest rather than on the command line
 
 ### `promote`
 
-> Behind the same feature gate as `config` and `up`.
-
 Make the version a workload is running permanent. The draft artifact it is serving is locked in place and given a version number; the workload keeps running it, and nothing is rebuilt or rolled.
 
 ```bash
@@ -381,8 +379,6 @@ Because `stop`, `start` and `delete` change something, they ask for confirmation
 
 ## Deploying from a project: `config` and `up`
 
-> These two, and `promote`, are alpha and behind a feature gate. Set `DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA=true` to see them in `--help`.
-
 `dr workload config` writes the `.datarobot.yaml` that describes your project, and `dr workload up` deploys the difference between that file and what is running. Together they are the deploy loop for a repository, where `create` deploys an artifact you already have.
 
 ```bash
@@ -480,7 +476,7 @@ dr workload delete <workload-id>
 
 ### Every command answers `403`
 
-A `403` on the very first request, including a plain `dr workload list`, is usually about access rather than about the workload you asked for. The Workload API sits behind a platform entitlement, and the CLI's own feature gate is unrelated to it: `DATAROBOT_CLI_FEATURE_WORKLOAD_ALPHA` only decides which commands the binary registers, never what your account may call.
+A `403` on the very first request, including a plain `dr workload list`, is usually about access rather than about the workload you asked for. The Workload API sits behind a platform entitlement, and the CLI's own feature gates are unrelated to it: a `DATAROBOT_CLI_FEATURE_*` variable only decides which commands the binary registers, never what your account may call.
 
 The response body names the check that refused you. The two common ones are a feature flag (a message naming `WORKLOAD_API_CONTAINERS`, or saying the Workload API is disabled in feature flags) and a seat license (a message saying the Agentic, Predictive and Governance seat has not been granted to this user). Both are grants your DataRobot administrator makes; neither can be worked around from the CLI. Entitlements are cached per user on the platform for a few minutes, so a newly granted one can keep failing briefly after it is switched on.