diff --git a/Project.toml b/Project.toml
index 18dafa1234..c67f713934 100644
--- a/Project.toml
+++ b/Project.toml
@@ -48,7 +48,7 @@ ChainRulesCore = "1.12"
 Compat = "4.10.0"
 Enzyme = "0.12, 0.13"
 Functors = "0.4"
-MLDataDevices = "1.4.0"
+MLDataDevices = "1.4.1"
 MLUtils = "0.4"
 MPI = "0.20.19"
 MacroTools = "0.5"
diff --git a/test/ext_cuda/cuda.jl b/test/ext_cuda/cuda.jl
index 8afba712fe..c2f3cefd2f 100644
--- a/test/ext_cuda/cuda.jl
+++ b/test/ext_cuda/cuda.jl
@@ -113,10 +113,10 @@ end
 
   # Even more trivial: no movement
   @test gradient(x -> sum(abs, cpu(x)), a)[1] isa Matrix
-  @test_broken gradient(x -> sum(abs, cpu(x)), a')[1] isa Matrix
+  @test gradient(x -> sum(abs, cpu(x)), a')[1] isa Matrix
   @test gradient(x -> sum(cpu(x)), a)[1] isa typeof(gradient(sum, a)[1]) # FillArray
   @test gradient(x -> sum(abs, gpu(x)), ca)[1] isa CuArray
-  @test_broken gradient(x -> sum(abs, gpu(x)), ca')[1] isa CuArray
+  @test gradient(x -> sum(abs, gpu(x)), ca')[1] isa CuArray
 
   # More complicated, Array * CuArray is an error
   g0 = gradient(x -> sum(abs, (a * (a * x))), a)[1]
@@ -131,8 +131,8 @@ end
 
   # Scalar indexing of an array, needs OneElement to transfer to GPU
   # https://github.com/FluxML/Zygote.jl/issues/1005
-  @test_broken gradient(x -> cpu(2 .* gpu(x))[1], Float32[1,2,3]) == ([2,0,0],)
-  @test_broken gradient(x -> cpu(gpu(x) * gpu(x))[1,2], Float32[1 2 3; 4 5 6; 7 8 9]) == ([2 6 8; 0 2 0; 0 3 0],)
+  @test gradient(x -> cpu(2 .* gpu(x))[1], Float32[1,2,3]) == ([2,0,0],)
+  @test gradient(x -> cpu(gpu(x) * gpu(x))[1,2], Float32[1 2 3; 4 5 6; 7 8 9]) == ([2 6 8; 0 2 0; 0 3 0],)
 end
 
 @testset "gpu(x) and cpu(x) on structured arrays" begin