diff --git a/pkg/cmd/create/create.go b/pkg/cmd/create/create.go
index 59b50979..e9fbfd32 100644
--- a/pkg/cmd/create/create.go
+++ b/pkg/cmd/create/create.go
@@ -69,7 +69,7 @@ func NewCmdCreate(t *terminal.Terminal, createStore CreateStore) *cobra.Command
 	}
 	cmd.Flags().BoolVarP(&detached, "detached", "d", false, "run the command in the background instead of blocking the shell")
 	cmd.Flags().StringVarP(&cpu, "cpu", "c", "", "CPU instance type. Defaults to 2x8 [2x8, 4x16, 8x32, 16x32]. See docs.brev.dev/cpu for details")
-	cmd.Flags().StringVarP(&gpu, "gpu", "g", "", "GPU instance type. See docs.brev.dev/gpu for details")
+	cmd.Flags().StringVarP(&gpu, "gpu", "g", "n1-highmem-4:nvidia-tesla-t4:1", "GPU instance type. See https://brev.dev/docs/reference/gpu for details")
 	return cmd
 }
 
diff --git a/pkg/cmd/start/start.go b/pkg/cmd/start/start.go
index 4d6bfb94..257abe98 100644
--- a/pkg/cmd/start/start.go
+++ b/pkg/cmd/start/start.go
@@ -29,7 +29,11 @@ var (
   brev start <git url>
   brev start <git url> --org myFancyOrg
 	`
-	instanceTypes = []string{"p4d.24xlarge", "p3.2xlarge", "p3.8xlarge", "p3.16xlarge", "p3dn.24xlarge", "p2.xlarge", "p2.8xlarge", "p2.16xlarge", "g5.xlarge", "g5.2xlarge", "g5.4xlarge", "g5.8xlarge", "g5.16xlarge", "g5.12xlarge", "g5.24xlarge", "g5.48xlarge", "g5g.xlarge", "g5g.2xlarge", "g5g.4xlarge", "g5g.8xlarge", "g5g.16xlarge", "g5g.metal", "g4dn.xlarge", "g4dn.2xlarge", "g4dn.4xlarge", "g4dn.8xlarge", "g4dn.16xlarge", "g4dn.12xlarge", "g4dn.metal", "g4ad.xlarge", "g4ad.2xlarge", "g4ad.4xlarge", "g4ad.8xlarge", "g4ad.16xlarge", "g3s.xlarge", "g3.4xlarge", "g3.8xlarge", "g3.16xlarge"}
+	instanceTypes = []string{
+		"p4d.24xlarge", "p3.2xlarge", "p3.8xlarge", "p3.16xlarge", "p3dn.24xlarge", "p2.xlarge", "p2.8xlarge", "p2.16xlarge", "g5.xlarge", "g5.2xlarge", "g5.4xlarge", "g5.8xlarge", "g5.16xlarge", "g5.12xlarge", "g5.24xlarge", "g5.48xlarge", "g5g.xlarge", "g5g.2xlarge", "g5g.4xlarge", "g5g.8xlarge", "g5g.16xlarge", "g5g.metal", "g4dn.xlarge", "g4dn.2xlarge", "g4dn.4xlarge", "g4dn.8xlarge", "g4dn.16xlarge", "g4dn.12xlarge", "g4dn.metal", "g4ad.xlarge", "g4ad.2xlarge", "g4ad.4xlarge", "g4ad.8xlarge", "g4ad.16xlarge", "g3s.xlarge", "g3.4xlarge", "g3.8xlarge", "g3.16xlarge",
+		"g5g.xlarge", "g5g.2xlarge", "g5g.4xlarge", "g5g.8xlarge", "g5g.16xlarge", "g5g.metal", "g4dn.xlarge", "g4dn.2xlarge", "g4dn.4xlarge", "g4dn.8xlarge", "g4dn.16xlarge", "g4dn.12xlarge", "g4dn.metal",
+		"n1-standard-1:nvidia-tesla-t4:1", "n1-highcpu-2:nvidia-tesla-t4:1", "n1-standard-2:nvidia-tesla-t4:1", "n1-highmem-2:nvidia-tesla-t4:1", "n1-highcpu-4:nvidia-tesla-t4:1", "n1-standard-4:nvidia-tesla-t4:1", "n1-highmem-4:nvidia-tesla-t4:1", "n1-highcpu-8:nvidia-tesla-t4:1", "n1-standard-1:nvidia-tesla-t4:2", "n1-highcpu-2:nvidia-tesla-t4:2", "n1-standard-8:nvidia-tesla-t4:1", "n1-standard-2:nvidia-tesla-t4:2", "n1-highmem-2:nvidia-tesla-t4:2", "n1-highcpu-4:nvidia-tesla-t4:2", "n1-highmem-8:nvidia-tesla-t4:1", "n1-standard-4:nvidia-tesla-t4:2", "n1-highmem-4:nvidia-tesla-t4:2", "n1-highcpu-16:nvidia-tesla-t4:1", "n1-highcpu-8:nvidia-tesla-t4:2", "n1-standard-8:nvidia-tesla-t4:2", "n1-standard-16:nvidia-tesla-t4:1", "n1-highmem-8:nvidia-tesla-t4:2", "n1-highcpu-16:nvidia-tesla-t4:2", "n1-standard-1:nvidia-tesla-t4:4", "n1-highcpu-2:nvidia-tesla-t4:4", "n1-highmem-16:nvidia-tesla-t4:1", "n1-standard-2:nvidia-tesla-t4:4", "n1-highmem-2:nvidia-tesla-t4:4", "n1-highcpu-4:nvidia-tesla-t4:4", "n1-standard-16:nvidia-tesla-t4:2", "n1-standard-4:nvidia-tesla-t4:4", "n1-highmem-4:nvidia-tesla-t4:4", "n1-highcpu-32:nvidia-tesla-t4:1", "n1-highcpu-8:nvidia-tesla-t4:4", "n1-highmem-16:nvidia-tesla-t4:2", "n1-standard-8:nvidia-tesla-t4:4", "n1-highmem-8:nvidia-tesla-t4:4", "n1-highcpu-32:nvidia-tesla-t4:2", "n1-highcpu-16:nvidia-tesla-t4:4", "n1-standard-32:nvidia-tesla-t4:1", "n1-standard-16:nvidia-tesla-t4:4", "n1-standard-32:nvidia-tesla-t4:2", "n1-highmem-16:nvidia-tesla-t4:4", "n1-highmem-32:nvidia-tesla-t4:1", "n1-highcpu-32:nvidia-tesla-t4:4", "n1-highmem-32:nvidia-tesla-t4:2", "n1-highcpu-64:nvidia-tesla-t4:1", "n1-standard-32:nvidia-tesla-t4:4", "n1-highcpu-64:nvidia-tesla-t4:2", "n1-highmem-32:nvidia-tesla-t4:4", "n1-standard-64:nvidia-tesla-t4:1", "n1-highcpu-64:nvidia-tesla-t4:4", "n1-standard-64:nvidia-tesla-t4:2", "n1-highcpu-96:nvidia-tesla-t4:1", "n1-highcpu-96:nvidia-tesla-t4:2", "n1-highmem-64:nvidia-tesla-t4:1", "n1-standard-64:nvidia-tesla-t4:4", "n1-highmem-64:nvidia-tesla-t4:2", "n1-highcpu-96:nvidia-tesla-t4:4", "n1-standard-96:nvidia-tesla-t4:1", "n1-highmem-64:nvidia-tesla-t4:4", "n1-standard-96:nvidia-tesla-t4:2", "n1-ultramem-40:nvidia-tesla-t4:1", "n1-standard-96:nvidia-tesla-t4:4", "n1-ultramem-40:nvidia-tesla-t4:2", "n1-highmem-96:nvidia-tesla-t4:1", "n1-highmem-96:nvidia-tesla-t4:2", "n1-ultramem-40:nvidia-tesla-t4:4", "n1-highmem-96:nvidia-tesla-t4:4", "n1-megamem-96:nvidia-tesla-t4:1", "n1-megamem-96:nvidia-tesla-t4:2", "n1-megamem-96:nvidia-tesla-t4:4", "n1-ultramem-80:nvidia-tesla-t4:1", "n1-ultramem-80:nvidia-tesla-t4:2", "n1-ultramem-80:nvidia-tesla-t4:4", "n1-ultramem-160:nvidia-tesla-t4:1", "n1-ultramem-160:nvidia-tesla-t4:2", "n1-ultramem-160:nvidia-tesla-t4:4",
+	}
 )
 
 type StartStore interface {
@@ -118,7 +122,7 @@ func NewCmdStart(t *terminal.Terminal, startStore StartStore, noLoginStartStore
 	cmd.Flags().StringVarP(&setupPath, "setup-path", "p", "", "path to env setup script. If you include --setup-repo we will apply this argument to that repo")
 	cmd.Flags().StringVarP(&org, "org", "o", "", "organization (will override active org if creating a workspace)")
 	// GPU options
-	cmd.Flags().StringVarP(&gpu, "gpu", "g", "", "GPU instance type. See docs.brev.dev/gpu for details")
+	cmd.Flags().StringVarP(&gpu, "gpu", "g", "n1-highmem-4:nvidia-tesla-t4:1", "GPU instance type. See https://brev.dev/docs/reference/gpu for details")
 	err := cmd.RegisterFlagCompletionFunc("org", completions.GetOrgsNameCompletionHandler(noLoginStartStore, t))
 	if err != nil {
 		breverrors.GetDefaultErrorReporter().ReportError(breverrors.WrapAndTrace(err))