diff --git a/muh/include/muh/tuning/tuning_histogram.cuh b/muh/include/muh/tuning/tuning_histogram.cuh
new file mode 100644
index 00000000..54611bfe
--- /dev/null
+++ b/muh/include/muh/tuning/tuning_histogram.cuh
@@ -0,0 +1,48 @@
+// muh/include/muh/tuning/tuning_histogram.cuh — BI-V100
+//
+// Mirrors: cccl_upstream/cub/cub/device/dispatch/tuning/tuning_histogram.cuh
+// CCCL SM100: 4 specializations by num_channels × num_active_channels
+//   Some use threads=1024 → SMEM overflow on BI-V100 for histogram bins > 48KB
+//
+// vllm relevance: token frequency histogram for repetition_penalty
+// SMEM risk: HIGH. Histogram SMEM = num_bins * sizeof(counter_t), NOT threads*items.
+//   256 bins * 4B = 1024B (safe). But 65536 bins → 256KB (overflow).
+//   BI-V100 limit: 49152 / 4 = 12288 bins max.
+
+#pragma once
+
+#include "muh/hardware.cuh"
+#include "muh/tuning/common.cuh"
+
+namespace muh::tuning::histogram {
+
+struct HistogramPolicy {
+  int threads_per_block;
+  int items_per_thread;
+  int privatized_bins_per_thread;
+  BlockLoadAlgorithm load_algorithm;
+  CacheLoadModifier load_modifier;
+  bool is_work_stealing;
+};
+
+struct policy_selector {
+  int num_channels;
+  int num_active_channels;
+  int max_bins;
+
+  constexpr HistogramPolicy operator()(const hardware_capability& hw) const {
+    // SMEM for privatized histogram: privatized_bins * sizeof(int) * threads/warp_size
+    // Must fit in 48KB
+    int max_privatized = hw.max_shared_memory_per_block / (4 * 8); // 8 = threads/warp conservative
+    int priv_bins = max_bins < max_privatized ? max_bins : max_privatized;
+    if (priv_bins < 1) priv_bins = 1;
+
+    if (num_channels == 1) {
+      return {384, 12, priv_bins, BLOCK_LOAD_DIRECT, LOAD_LDG, false};
+    }
+    // multi-channel: fewer threads to leave SMEM for bins
+    return {256, 8, priv_bins, BLOCK_LOAD_DIRECT, LOAD_LDG, false};
+  }
+};
+
+} // namespace muh::tuning::histogram