diff --git a/examples/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling.cu b/examples/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling.cu
index 94eb5532..bfa27370 100644
--- a/examples/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling.cu
+++ b/examples/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling.cu
@@ -738,7 +738,7 @@ int run(OptionType &options, bool host_problem_shapes_available = true)
 
 int main(int argc, char const **args) {
 
-  // CUTLASS must be compiled with CUDA 12.0 Toolkit to run this example
+  // CUTLASS must be compiled with CUDA 12.3 Toolkit to run this example
   // and must have compute capability at least 90.
   if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 3)) {
     std::cerr << "This example requires CUDA 12.3 or newer.\n";
diff --git a/examples/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling_with_sparse_groups.cu b/examples/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling_with_sparse_groups.cu
index b5419fe2..a43af589 100644
--- a/examples/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling_with_sparse_groups.cu
+++ b/examples/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling/68_hopper_fp8_warp_specialized_grouped_gemm_with_blockwise_scaling_with_sparse_groups.cu
@@ -751,7 +751,7 @@ int run(OptionType &options, bool host_problem_shapes_available = true)
 
 int main(int argc, char const **args) {
 
-  // CUTLASS must be compiled with CUDA 12.0 Toolkit to run this example
+  // CUTLASS must be compiled with CUDA 12.3 Toolkit to run this example
   // and must have compute capability at least 90.
   if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 3)) {
     std::cerr << "This example requires CUDA 12.3 or newer.\n";
diff --git a/examples/70_blackwell_gemm/70_blackwell_fp16_gemm.cu b/examples/70_blackwell_gemm/70_blackwell_fp16_gemm.cu
index f9d14215..bcbe9712 100644
--- a/examples/70_blackwell_gemm/70_blackwell_fp16_gemm.cu
+++ b/examples/70_blackwell_gemm/70_blackwell_fp16_gemm.cu
@@ -440,7 +440,7 @@ int run(Options &options)
 
 int main(int argc, char const **args) {
 
-  // CUTLASS must be compiled with CUDA 12.0 Toolkit to run this example
+  // CUTLASS must be compiled with CUDA 12.8 Toolkit to run this example
   // and must have compute capability at least 100a.
 
   if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 8)) {
diff --git a/examples/70_blackwell_gemm/70_blackwell_fp8_gemm.cu b/examples/70_blackwell_gemm/70_blackwell_fp8_gemm.cu
index f0b85865..6ab0fab5 100644
--- a/examples/70_blackwell_gemm/70_blackwell_fp8_gemm.cu
+++ b/examples/70_blackwell_gemm/70_blackwell_fp8_gemm.cu
@@ -626,11 +626,11 @@ int run(Options &options)
 
 int main(int argc, char const **args) {
 
-  // CUTLASS must be compiled with CUDA 12.0 Toolkit to run this example
+  // CUTLASS must be compiled with CUDA 12.8 Toolkit to run this example
   // and must have compute capability at least sm100a.
-  
-  if (__CUDACC_VER_MAJOR__ < 12) {
-    std::cerr << "This example requires CUDA 12 or newer.\n";
+
+  if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 8)) {
+    std::cerr << "This example requires CUDA 12.8 or newer." << std::endl;
     // Returning zero so this test passes on older Toolkits. Its actions are no-op.
     return 0;
   }
diff --git a/examples/76_blackwell_conv/76_blackwell_conv_wgrad.cu b/examples/76_blackwell_conv/76_blackwell_conv_wgrad.cu
index a491bed8..ff4cac19 100644
--- a/examples/76_blackwell_conv/76_blackwell_conv_wgrad.cu
+++ b/examples/76_blackwell_conv/76_blackwell_conv_wgrad.cu
@@ -486,8 +486,8 @@ int run(Options &options)
 
 int main(int argc, char const **args) {
 
-  // CUTLASS must be compiled with CUDA 12.0 Toolkit to run this example
-  // and must have compute capability at least 90.
+  // CUTLASS must be compiled with CUDA 12.8 Toolkit to run this example
+  // and must have compute capability at least sm100a.
   if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 8)) {
     std::cerr << "This example requires CUDA 12.8 or newer." << std::endl;
     // Returning zero so this test passes on older Toolkits. Its actions are no-op.
diff --git a/examples/77_blackwell_fmha/77_blackwell_fmha.cu b/examples/77_blackwell_fmha/77_blackwell_fmha.cu
index cc09e68a..18ffd7fe 100644
--- a/examples/77_blackwell_fmha/77_blackwell_fmha.cu
+++ b/examples/77_blackwell_fmha/77_blackwell_fmha.cu
@@ -1036,10 +1036,16 @@ int main_single(int argc, char const **args) {
     return -1;
   }
 
-  if (__CUDACC_VER_MAJOR__ < 12 || props.major != 10) {
-    std::cout
+  if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 8)) {
+    std::cerr << "This example requires CUDA 12.8 or newer." << std::endl;
+    // Returning zero so this test passes on older Toolkits. Its actions are no-op.
+    return 0;
+  }
+
+  if (props.major != 10) {
+    std::cerr
       << "This example requires a GPU of NVIDIA's Blackwell Architecture "
-      << "(compute capability major 10) and CUDA 12.8 or greater.\n";
+      << "(compute capability 100a)." << std::endl;
     return 0;
   }
 
diff --git a/examples/77_blackwell_fmha/77_blackwell_fmha_bwd.cu b/examples/77_blackwell_fmha/77_blackwell_fmha_bwd.cu
index 4521d87f..75729d2e 100644
--- a/examples/77_blackwell_fmha/77_blackwell_fmha_bwd.cu
+++ b/examples/77_blackwell_fmha/77_blackwell_fmha_bwd.cu
@@ -939,10 +939,16 @@ int main_single(int argc, char const **args) {
     return -1;
   }
 
-  if (__CUDACC_VER_MAJOR__ < 12 || props.major != 10) {
-    std::cout
+  if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 8)) {
+    std::cerr << "This example requires CUDA 12.8 or newer." << std::endl;
+    // Returning zero so this test passes on older Toolkits. Its actions are no-op.
+    return 0;
+  }
+
+  if (props.major != 10) {
+    std::cerr
       << "This example requires a GPU of NVIDIA's Blackwell Architecture "
-      << "(compute capability 100a) and CUDA 12.8 or greater.\n";
+      << "(compute capability 100a)." << std::endl;
     return 0;
   }
   
diff --git a/examples/77_blackwell_fmha/77_blackwell_fmha_gen.cu b/examples/77_blackwell_fmha/77_blackwell_fmha_gen.cu
index 87ee6ac3..889d5d2b 100644
--- a/examples/77_blackwell_fmha/77_blackwell_fmha_gen.cu
+++ b/examples/77_blackwell_fmha/77_blackwell_fmha_gen.cu
@@ -722,12 +722,19 @@ int main_single(int argc, char const **args) {
     return -1;
   }
 
-  if (__CUDACC_VER_MAJOR__ < 12 || props.major < 10) {
-    std::cout
-      << "This example requires a GPU of NVIDIA's Blackwell Architecture or "
-      << "later (compute capability 90 or greater) and CUDA 12.0 or greater.\n";
+  if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 8)) {
+    std::cerr << "This example requires CUDA 12.8 or newer." << std::endl;
+    // Returning zero so this test passes on older Toolkits. Its actions are no-op.
     return 0;
   }
+
+  if (props.major != 10) {
+    std::cerr
+      << "This example requires a GPU of NVIDIA's Blackwell Architecture "
+      << "(compute capability 100a)." << std::endl;
+    return 0;
+  }
+
   //
   // Parse options
   //
diff --git a/examples/77_blackwell_fmha/77_blackwell_mla.cu b/examples/77_blackwell_fmha/77_blackwell_mla.cu
index fda0cebd..8c7ac410 100644
--- a/examples/77_blackwell_fmha/77_blackwell_mla.cu
+++ b/examples/77_blackwell_fmha/77_blackwell_mla.cu
@@ -766,10 +766,16 @@ int main_single(int argc, char const **args) {
     return -1;
   }
 
-  if (__CUDACC_VER_MAJOR__ < 12 || props.major != 10) {
-    std::cout
+  if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 8)) {
+    std::cerr << "This example requires CUDA 12.8 or newer." << std::endl;
+    // Returning zero so this test passes on older Toolkits. Its actions are no-op.
+    return 0;
+  }
+
+  if (props.major != 10) {
+    std::cerr
       << "This example requires a GPU of NVIDIA's Blackwell Architecture "
-      << "(compute capability major 10) and CUDA 12.8 or greater.\n";
+      << "(compute capability 100a)." << std::endl;
     return 0;
   }
 
diff --git a/examples/77_blackwell_fmha/77_blackwell_mla_fwd.cu b/examples/77_blackwell_fmha/77_blackwell_mla_fwd.cu
index 2de135a0..80a637b2 100644
--- a/examples/77_blackwell_fmha/77_blackwell_mla_fwd.cu
+++ b/examples/77_blackwell_fmha/77_blackwell_mla_fwd.cu
@@ -972,10 +972,16 @@ int main_single(int argc, char const **args) {
     return -1;
   }
 
-  if (__CUDACC_VER_MAJOR__ < 12 || props.major != 10) {
-    std::cout
+  if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 8)) {
+    std::cerr << "This example requires CUDA 12.8 or newer." << std::endl;
+    // Returning zero so this test passes on older Toolkits. Its actions are no-op.
+    return 0;
+  }
+
+  if (props.major != 10) {
+    std::cerr
       << "This example requires a GPU of NVIDIA's Blackwell Architecture "
-      << "(compute capability major 10) and CUDA 12.8 or greater.\n";
+      << "(compute capability 100a)." << std::endl;
     return 0;
   }
 
diff --git a/examples/78_blackwell_emulated_bf16x9_gemm/78_blackwell_emulated_bf16x9_gemm.cu b/examples/78_blackwell_emulated_bf16x9_gemm/78_blackwell_emulated_bf16x9_gemm.cu
index cd4231c0..880f114a 100644
--- a/examples/78_blackwell_emulated_bf16x9_gemm/78_blackwell_emulated_bf16x9_gemm.cu
+++ b/examples/78_blackwell_emulated_bf16x9_gemm/78_blackwell_emulated_bf16x9_gemm.cu
@@ -429,7 +429,7 @@ int run(Options &options)
 
 int main(int argc, char const **args) {
 
-  // CUTLASS must be compiled with CUDA 12.0 Toolkit to run this example
+  // CUTLASS must be compiled with CUDA 12.8 Toolkit to run this example
   // and must have compute capability at least 100a.
   if (__CUDACC_VER_MAJOR__ < 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ < 8)) {
     std::cerr << "This example requires CUDA 12.8 or newer." << std::endl;