diff --git a/jax/_src/pallas/mosaic/lowering.py b/jax/_src/pallas/mosaic/lowering.py
index b4863d196fc3..37d17464d46e 100644
--- a/jax/_src/pallas/mosaic/lowering.py
+++ b/jax/_src/pallas/mosaic/lowering.py
@@ -1058,11 +1058,7 @@ def _transpose_lowering_rule(ctx: LoweringRuleContext, x, *, permutation):
   if permutation != (1, 0):
     raise NotImplementedError
   out_type = aval_to_ir_type(ctx.avals_out[0])
-  i64_type = ir.IntegerType.get_signless(64)
-  transp = ir.ArrayAttr.get(
-      [ir.IntegerAttr.get(i64_type, i) for i in permutation]
-  )
-  return vector.TransposeOp(out_type, x, transp).result
+  return vector.TransposeOp(out_type, x, permutation).result
 
 
 lowering_rules[lax.transpose_p] = _transpose_lowering_rule
diff --git a/jaxlib/mosaic/dialect/tpu/transforms/apply_vector_layout.cc b/jaxlib/mosaic/dialect/tpu/transforms/apply_vector_layout.cc
index 091c9d9ea132..47105456b40d 100644
--- a/jaxlib/mosaic/dialect/tpu/transforms/apply_vector_layout.cc
+++ b/jaxlib/mosaic/dialect/tpu/transforms/apply_vector_layout.cc
@@ -2809,11 +2809,8 @@ LogicalResult vector_transpose_rule(RewriteContext &ctx, Operation &op,
       xla::Array<Value> src_vregs,
       disassemble(builder, layout_in, transpose_op.getVector(),
                   ctx.target_shape));
-  const SmallVector<int64_t> permutation =
-      llvm::map_to_vector(transpose_op.getTransp(), [&](const Attribute attr) {
-        return cast<IntegerAttr>(attr).getValue().getSExtValue();
-      });
-  const auto tile_perm = ArrayRef<int64_t>(permutation).take_back(2);
+  ArrayRef<int64_t> permutation = transpose_op.getPermutation();
+  const auto tile_perm = permutation.take_back(2);
   if (tile_perm != ArrayRef<int64_t>{rank - 2, rank - 1} &&
       tile_perm != ArrayRef<int64_t>{rank - 1, rank - 2}) {
     return transpose_op->emitOpError(
@@ -2851,8 +2848,7 @@ LogicalResult vector_transpose_rule(RewriteContext &ctx, Operation &op,
   const int packing = layout_in.packing();
   // Note that we checked for native tiling above.
   const int64_t vregs_per_tile = transpose_unit_size / layout_in.tiling()[0];
-  const ArrayAttr minor_perm = builder.getArrayAttr(
-      {builder.getI64IntegerAttr(1), builder.getI64IntegerAttr(0)});
+  const SmallVector<int64_t> minor_perm{1, 0};
   const auto tile_ty = VectorType::get(
       {transpose_unit_size, transpose_unit_size}, src_ty.getElementType());
   const auto batch_tile_ty_in =
diff --git a/jaxlib/mosaic/dialect/tpu/transforms/infer_vector_layout.cc b/jaxlib/mosaic/dialect/tpu/transforms/infer_vector_layout.cc
index 69327451e8de..85dc5d8ef665 100644
--- a/jaxlib/mosaic/dialect/tpu/transforms/infer_vector_layout.cc
+++ b/jaxlib/mosaic/dialect/tpu/transforms/infer_vector_layout.cc
@@ -413,11 +413,10 @@ class VectorLayoutInferer {
         }
         if (auto transpose =
                 dyn_cast<vector::TransposeOp>(operand.getOwner())) {
-          auto perm_attrs = transpose.getTransp().getValue();
-          auto rank = perm_attrs.size();
-          if (rank >= 2 &&
-              cast<IntegerAttr>(perm_attrs[rank - 1]).getInt() == rank - 2 &&
-              cast<IntegerAttr>(perm_attrs[rank - 2]).getInt() == rank - 1) {
+          auto perm = transpose.getPermutation();
+          auto rank = perm.size();
+          if (rank >= 2 && perm[rank - 1] == rank - 2 &&
+              perm[rank - 2] == rank - 1) {
             continue;
           }
           // Fall through.
@@ -1321,12 +1320,12 @@ class VectorLayoutInferer {
   }
 
   LogicalResult infer(vector::TransposeOp op) {
-    auto permutation_attrs = op.getTransp().getValue();
+    auto permutation = op.getPermutation();
     auto some_layout = getLayout(op.getVector());
     TPU_CHECK_OP(some_layout.has_value(), "missing vector layout");
     auto &layout = *some_layout;
     auto src_ty = op.getSourceVectorType();
-    TPU_CHECK_OP(permutation_attrs.size() == src_ty.getRank(),
+    TPU_CHECK_OP(permutation.size() == src_ty.getRank(),
                  "Transpose permutation has incorrect rank");
     if (layout.implicit_dim() == ImplicitDim::kNone) {
       TPU_CHECK_OP((layout.offsets() == LayoutOffsets{0, 0}),
@@ -1335,23 +1334,22 @@ class VectorLayoutInferer {
       for (int64_t s : src_ty.getShape().take_back(2)) {
         TPU_CHECK_OP(s % xlu_width == 0, "Padded transposes unsupported");
       }
-      for (auto attr : permutation_attrs.drop_back(2)) {
+      for (auto dim : permutation.drop_back(2)) {
         TPU_CHECK_OP(
-            cast<IntegerAttr>(attr).getInt() < src_ty.getRank() - 2,
+            dim < src_ty.getRank() - 2,
             "Unsupported transpose permutation - minor dims into major");
       }
-      for (auto attr : permutation_attrs.take_back(2)) {
+      for (auto dim : permutation.take_back(2)) {
         TPU_CHECK_OP(
-            cast<IntegerAttr>(attr).getInt() >= src_ty.getRank() - 2,
+            dim >= src_ty.getRank() - 2,
             "Unsupported transpose permutation - major dims into minor");
       }
       Layout required_layout = some_layout;
-      if (permutation_attrs.size() < 2) {
+      if (permutation.size() < 2) {
         return failure();
       }
       // Require native tiling if we're going to use the XLU.
-      if (cast<IntegerAttr>(permutation_attrs[permutation_attrs.size() - 1])
-              .getInt() == permutation_attrs.size() - 2) {
+      if (permutation[permutation.size() - 1] == permutation.size() - 2) {
         auto native_tiling = nativeTiling(layout.bitwidth());
         required_layout = VectorLayout(layout.bitwidth(), layout.offsets(),
                                        native_tiling, ImplicitDim::kNone);
diff --git a/jaxlib/mosaic/python/apply_vector_layout.py b/jaxlib/mosaic/python/apply_vector_layout.py
index 01ab44a24ad0..a4928bcfa810 100644
--- a/jaxlib/mosaic/python/apply_vector_layout.py
+++ b/jaxlib/mosaic/python/apply_vector_layout.py
@@ -3193,10 +3193,7 @@ def _vector_transpose_rule(  # pylint: disable=missing-function-docstring
   dst_ty = ir.VectorType(op.result.type)
   rank = src_ty.rank
   src_vregs = disassemble(layout_in, op.vector)
-  permutation = [
-      ir.IntegerAttr(attr).value
-      for attr in ir.ArrayAttr(op.attributes["transp"])
-  ]
+  permutation = [i for i in ir.DenseI64ArrayAttr(op.attributes["permutation"])]
   batch_perm, tile_perm = permutation[:-2], permutation[-2:]
   if set(batch_perm) != set(range(len(batch_perm))):
     raise NotImplementedError("Unsupported major permutation")
@@ -3217,7 +3214,7 @@ def _vector_transpose_rule(  # pylint: disable=missing-function-docstring
   packing = layout_in.packing
   # Note that we checked for native tiling above.
   vregs_per_tile = transpose_unit_size // layout_in.tiling[0]
-  minor_perm = ir.ArrayAttr.get([ir.IntegerAttr.get(i64(), i) for i in (1, 0)])
+  minor_perm = [1, 0]
   tile_ty = ir.VectorType.get((transpose_unit_size,) * 2, src_ty.element_type)
   batch_tile_ty_in = ir.VectorType.get(
       (transpose_unit_size, transpose_unit_size * packing), src_ty.element_type