diff --git a/CHANGELOG.md b/CHANGELOG.md
index b6c0f6b..00ac862 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,6 +2,18 @@
 
 All notable changes to this project are documented in this file.
 
+## [1.2.0] - 2025-12-20
+### Changed
+- `reverse/1` now uses the BEAM stdlib grapheme segmentation (`string.to_graphemes`) for better Unicode correctness and consistency across the library.
+
+### Performance
+- Optimized `count/3` internals to avoid repeated `list.length` calls inside recursive loops, improving performance on long strings.
+
+### CI
+- Pinned Gleam version in CI and fixed build cache path/key for more reproducible and faster runs.
+
+Contributed by: Daniele (`lupodevelop`)
+
 ## [1.1.1] - 2025-11-30
 ### Fixed
 - Robustness fixes for grapheme-aware utilities; resolved parity issues in `ends_with/2` for complex ZWJ sequences.
diff --git a/gleam.toml b/gleam.toml
index 773d060..dc062b1 100644
--- a/gleam.toml
+++ b/gleam.toml
@@ -1,5 +1,5 @@
 name = "str"
-version = "1.1.1"
+version = "1.2.0"
 
 # Project metadata (fill or replace placeholders before publishing)
 description = "Unicode-aware string utilities for Gleam: grapheme-safe operations, pragmatic ASCII transliteration, and slug generation."
diff --git a/src/str/core.gleam b/src/str/core.gleam
index 127a623..4c395f5 100644
--- a/src/str/core.gleam
+++ b/src/str/core.gleam
@@ -12,7 +12,6 @@
 import gleam/int
 import gleam/list
 import gleam/string
-import str/tokenize
 
 /// Detects if a grapheme cluster likely contains emoji components.
 ///
@@ -195,24 +194,51 @@ fn count_loop(
   overlapping: Bool,
   acc: Int,
 ) -> Int {
-  case list.length(hs) < nd_len {
+  count_loop_with_len(hs, list.length(hs), nd, nd_len, overlapping, acc)
+}
+
+fn count_loop_with_len(
+  hs: List(String),
+  hs_len: Int,
+  nd: List(String),
+  nd_len: Int,
+  overlapping: Bool,
+  acc: Int,
+) -> Int {
+  case hs_len < nd_len {
     True -> acc
     False ->
       case list.take(hs, nd_len) == nd {
         True ->
           case overlapping {
             True ->
-              count_loop(list.drop(hs, 1), nd, nd_len, overlapping, acc + 1)
+              count_loop_with_len(
+                list.drop(hs, 1),
+                hs_len - 1,
+                nd,
+                nd_len,
+                overlapping,
+                acc + 1,
+              )
             False ->
-              count_loop(
+              count_loop_with_len(
                 list.drop(hs, nd_len),
+                hs_len - nd_len,
                 nd,
                 nd_len,
                 overlapping,
                 acc + 1,
               )
           }
-        False -> count_loop(list.drop(hs, 1), nd, nd_len, overlapping, acc)
+        False ->
+          count_loop_with_len(
+            list.drop(hs, 1),
+            hs_len - 1,
+            nd,
+            nd_len,
+            overlapping,
+            acc,
+          )
       }
   }
 }
@@ -413,10 +439,10 @@ pub fn truncate_default(text: String, max_len: Int) -> String {
 ///   reverse("👨‍👩‍👧‍👦") -> "👨‍👩‍👧‍👦"
 ///
 pub fn reverse(text: String) -> String {
-  let clusters = tokenize.chars(text)
-  clusters
+  text
+  |> string.to_graphemes
   |> list.reverse
-  |> list.fold("", fn(acc, s) { acc <> s })
+  |> string.concat
 }
 
 // ============================================================================