diff --git a/.nextchanges/air/accelerator-field-compute-help.md b/.nextchanges/air/accelerator-field-compute-help.md
new file mode 100644
index 00000000000..bdc6b13b9e0
--- /dev/null
+++ b/.nextchanges/air/accelerator-field-compute-help.md
@@ -0,0 +1 @@
+* Show workspace-supported accelerator types in `databricks air run -h config.compute.accelerator_type`. ([#7275](https://github.com/databricks/cli/pull/7275))
diff --git a/acceptance/air/config-help/output.txt b/acceptance/air/config-help/output.txt
index 16fedb9ffa9..4df01151206 100644
--- a/acceptance/air/config-help/output.txt
+++ b/acceptance/air/config-help/output.txt
@@ -75,7 +75,7 @@ config.compute
   Which accelerators to run on and how many.
 
   Supported accelerator types:
-    GPU_1xH100 - 1 accelerator per node [Private Preview]
+    GPU_1xH100 - 1 accelerator per node
     GPU_8xH100 - 8 accelerators per node
     GPU_1xRTX6000 - 1 accelerator per node
 
@@ -128,6 +128,11 @@ config.environment.version
 config.compute.accelerator_type
   Which accelerator to run on, e.g. GPU_1xA10. See https://docs.databricks.com/aws/en/machine-learning/ai-runtime/cli/yaml-config#reference for the current list of supported GPU types. Matched case-sensitively.
 
+  Supported accelerator types:
+    GPU_1xH100 - 1 accelerator per node
+    GPU_8xH100 - 8 accelerators per node
+    GPU_1xRTX6000 - 1 accelerator per node
+
   Type:     string
   Required: no
 
diff --git a/acceptance/air/config-help/test.toml b/acceptance/air/config-help/test.toml
index 79cb5772cdf..6759bc32089 100644
--- a/acceptance/air/config-help/test.toml
+++ b/acceptance/air/config-help/test.toml
@@ -9,7 +9,7 @@ Pattern = "GET /api/2.0/ai-training/compute-options"
 Response.Body = '''
 {
   "compute_options": [
-    {"hardware_accelerator": "GPU_1xH100", "per_node_accelerator_count": 1, "launch_stage": "PRIVATE_PREVIEW"},
+    {"hardware_accelerator": "GPU_1xH100", "per_node_accelerator_count": 1, "launch_stage": "GA"},
     {"hardware_accelerator": "GPU_8xH100", "per_node_accelerator_count": 8, "launch_stage": "GA"},
     {"hardware_accelerator": "GPU_1xRTX6000", "per_node_accelerator_count": 1, "launch_stage": "GA"}
   ]
diff --git a/cmd/air/run.go b/cmd/air/run.go
index d63193e5888..332f1aa1218 100644
--- a/cmd/air/run.go
+++ b/cmd/air/run.go
@@ -122,7 +122,7 @@ The path must be a separate argument: cobra reserves -h as a boolean, so
 		}
 
 		var computeOptions []computeOption
-		if field.path == "config.compute" {
+		if field.path == "config.compute" || field.path == "config.compute.accelerator_type" {
 			computeOptions = loadComputeOptionsForHelp(c, args)
 		}
 		renderConfigField(c.OutOrStdout(), field, computeOptions)
diff --git a/cmd/air/runconfig.go b/cmd/air/runconfig.go
index 31c701cdc12..e4b8fb5411e 100644
--- a/cmd/air/runconfig.go
+++ b/cmd/air/runconfig.go
@@ -862,7 +862,7 @@ func renderConfigField(w io.Writer, f configField, computeOptions []computeOptio
 	if f.help != "" {
 		fmt.Fprintf(w, "  %s\n", f.help)
 	}
-	if f.path == "config.compute" && computeOptions != nil {
+	if (f.path == "config.compute" || f.path == "config.compute.accelerator_type") && computeOptions != nil {
 		fmt.Fprintln(w, "\n  Supported accelerator types:")
 		if len(computeOptions) == 0 {
 			fmt.Fprintln(w, "    No accelerator types are currently available for AI Runtime in this workspace.")
diff --git a/cmd/air/runconfig_test.go b/cmd/air/runconfig_test.go
index e3c2f34520a..e2d4b47c934 100644
--- a/cmd/air/runconfig_test.go
+++ b/cmd/air/runconfig_test.go
@@ -984,12 +984,19 @@ func TestRunCommandHelp_NoConfigPath(t *testing.T) {
 // -h with a config path documents the field instead of the command, and does so
 // without the otherwise-required --file.
 func TestRunCommandHelp_ConfigPath(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		_, _ = w.Write([]byte(`{"compute_options":[{"hardware_accelerator":"GPU_FUTURE","per_node_accelerator_count":16}]}`))
+	}))
+	t.Cleanup(srv.Close)
+
 	var out strings.Builder
 	cmd := newRunCommand()
+	cmd.SetContext(cmdctx.SetWorkspaceClient(t.Context(), newTestWorkspaceClient(t, srv.URL)))
 	cmd.SetOut(&out)
 	cmd.SetArgs([]string{"-h", "config.compute.accelerator_type"})
 	require.NoError(t, cmd.Execute())
 	assert.Contains(t, out.String(), "config.compute.accelerator_type")
+	assert.Contains(t, out.String(), "GPU_FUTURE - 16 accelerators per node")
 	assert.Contains(t, out.String(), "Type:     string")
 	assert.NotContains(t, out.String(), "Usage:")
 }