diff --git a/doc/changes/fixed/12869.md b/doc/changes/fixed/12869.md
new file mode 100644
index 00000000000..cb27912341b
--- /dev/null
+++ b/doc/changes/fixed/12869.md
@@ -0,0 +1,7 @@
+- Fix greedy version location in lang declarations. Previously, error locations for
+  invalid lang versions would span multiple bytes for multi-byte UTF-8 characters,
+  causing carets to appear misaligned and seemingly include the closing
+  parenthesis. Now, error locations for ASCII strings show the full length (e.g.,
+  "Ali" shows `^^^`), while non-ASCII strings show only the first byte (e.g., "è"
+  shows `^`) to avoid multi-byte character display issues. (#12869, fixes #12806,
+  @benodiwal)
diff --git a/src/dune_sexp/versioned_file_first_line.mll b/src/dune_sexp/versioned_file_first_line.mll
index e5459b3d079..2e5bac4a96c 100644
--- a/src/dune_sexp/versioned_file_first_line.mll
+++ b/src/dune_sexp/versioned_file_first_line.mll
@@ -9,11 +9,18 @@ let invalid_lang_line start lexbuf =
   lexbuf.Lexing.lex_start_p <- start;
   User_error.raise ~loc:(Loc.of_lexbuf lexbuf)
     [ Pp.text "Invalid first line, expected: (lang <lang> <version>)" ]
+
+let rec find_first_non_ascii s i =
+  if i >= String.length s then None
+  else if Char.code s.[i] >= 128 then Some i
+  else find_first_non_ascii s (i + 1)
 }
 
 let newline   = '\r'? '\n'
 let blank     = [' ' '\t']
-let atom_char = [^';' '(' ')' '"' '#' '|' '\000'-'\032']
+let ascii = [ '\000' - '\127' ]
+let atom_char = [^ ';' '(' ')' '"' '\000'-'\032' '\127'-'\255']
+let atom_char_with_junk = [^';' '(' ')' '"' '#' '|' '\000'-'\032'] | [ '\127'-'\255' ]
 
 rule lex_opt = parse
   | '(' blank* "lang"
@@ -34,6 +41,22 @@ and atom start = parse
   | atom_char+ as s
     { (Loc.of_lexbuf lexbuf, s)
     }
+  | atom_char_with_junk+ as s
+    { let atom_start = Lexing.lexeme_start_p lexbuf in
+      (* CR-someday benodiwal: Consider using Uuseg for more accurate grapheme
+         cluster detection instead of the simple byte-based check. This would
+         better handle complex Unicode like emoji sequences, though terminal
+         rendering inconsistencies would still cause display issues. *)
+      let loc_start, loc_end =
+        match find_first_non_ascii s 0 with
+        | None -> assert false
+        | Some offset ->
+          let error_start = { atom_start with pos_cnum = atom_start.pos_cnum + offset } in
+          let error_end = { error_start with pos_cnum = error_start.pos_cnum + 1 } in
+          error_start, error_end
+      in
+      (Loc.create ~start:loc_start ~stop:loc_end, s)
+    }
   | _ | eof
     { to_eol lexbuf;
       invalid_lang_line start lexbuf
diff --git a/test/blackbox-tests/test-cases/lang-invalid-version.t b/test/blackbox-tests/test-cases/lang-invalid-version.t
index d53a20c8e7c..2bdd6af3b84 100644
--- a/test/blackbox-tests/test-cases/lang-invalid-version.t
+++ b/test/blackbox-tests/test-cases/lang-invalid-version.t
@@ -24,48 +24,78 @@ atoms without needing validation in every decoder.
 
 Test with various non-ASCII characters:
 
-CR-someday benodiwal: The version_loc is greedy and captures the closing
-parenthesis.
-
   $ test_invalid_version "è"
-  File "dune-project", line 1, characters 11-13:
+  File "dune-project", line 1, characters 11-12:
   1 | (lang dune è)
-                 ^^
+                 ^
   Error: Invalid atom: contains non-ASCII character(s). Atoms must only contain
   ASCII characters.
   [1]
 
   $ test_invalid_version "π3.14"
-  File "dune-project", line 1, characters 11-17:
+  File "dune-project", line 1, characters 11-12:
   1 | (lang dune π3.14)
-                 ^^^^^^
+                 ^
   Error: Invalid atom: contains non-ASCII character(s). Atoms must only contain
   ASCII characters.
   [1]
 
   $ test_invalid_version "α"
-  File "dune-project", line 1, characters 11-13:
+  File "dune-project", line 1, characters 11-12:
   1 | (lang dune α)
-                 ^^
+                 ^
   Error: Invalid atom: contains non-ASCII character(s). Atoms must only contain
   ASCII characters.
   [1]
 
   $ test_invalid_version "😀"
-  File "dune-project", line 1, characters 11-15:
+  File "dune-project", line 1, characters 11-12:
   1 | (lang dune 😀)
-                 ^^^^
+                 ^
   Error: Invalid atom: contains non-ASCII character(s). Atoms must only contain
   ASCII characters.
   [1]
 
-CR-someday benodiwal: Unicode string lengths are miscomputed in location
-excerpts for East Asian characters.
-
   $ test_invalid_version "中3.16文"
-  File "dune-project", line 1, characters 11-21:
+  File "dune-project", line 1, characters 11-12:
   1 | (lang dune 中3.16文)
-                 ^^^^^^^^^^
+                 ^
+  Error: Invalid atom: contains non-ASCII character(s). Atoms must only contain
+  ASCII characters.
+  [1]
+
+Test where non-ASCII character is not the first character:
+
+  $ test_invalid_version "3.14è"
+  File "dune-project", line 1, characters 15-16:
+  1 | (lang dune 3.14è)
+                     ^
+  Error: Invalid atom: contains non-ASCII character(s). Atoms must only contain
+  ASCII characters.
+  [1]
+
+  $ test_invalid_version "2è3"
+  File "dune-project", line 1, characters 12-13:
+  1 | (lang dune 2è3)
+                  ^
+  Error: Invalid atom: contains non-ASCII character(s). Atoms must only contain
+  ASCII characters.
+  [1]
+
+  $ test_invalid_version "abc😀def"
+  File "dune-project", line 1, characters 14-15:
+  1 | (lang dune abc😀def)
+                    ^
+  Error: Invalid atom: contains non-ASCII character(s). Atoms must only contain
+  ASCII characters.
+  [1]
+
+Test with multiple non-ASCII characters (caret points to first one):
+
+  $ test_invalid_version "aè😀"
+  File "dune-project", line 1, characters 12-13:
+  1 | (lang dune aè😀)
+                  ^
   Error: Invalid atom: contains non-ASCII character(s). Atoms must only contain
   ASCII characters.
   [1]