diff --git a/Misc/NEWS.d/next/Core_and_Builtins/2026-07-11-14-57-19.gh-issue-153568.memoflt.rst b/Misc/NEWS.d/next/Core_and_Builtins/2026-07-11-14-57-19.gh-issue-153568.memoflt.rst
new file mode 100644
index 000000000000000..9e78e5ad545a83a
--- /dev/null
+++ b/Misc/NEWS.d/next/Core_and_Builtins/2026-07-11-14-57-19.gh-issue-153568.memoflt.rst
@@ -0,0 +1,2 @@
+Speed up the parser by letting memoization lookups that cannot match return
+immediately.
diff --git a/Parser/pegen.c b/Parser/pegen.c
index 165dcb9f9950955..df83469895e03c0 100644
--- a/Parser/pegen.c
+++ b/Parser/pegen.c
@@ -90,6 +90,7 @@ _PyPegen_insert_memo(Parser *p, int mark, int type, void *node)
     m->mark = p->mark;
     m->next = p->tokens[mark]->memo;
     p->tokens[mark]->memo = m;
+    p->tokens[mark]->memo_mask |= 1ULL << (type & 63);
     return 0;
 }
 
@@ -350,6 +351,10 @@ _PyPegen_is_memoized(Parser *p, int type, void *pres)
 
     Token *t = p->tokens[p->mark];
 
+    if (!(t->memo_mask & (1ULL << (type & 63)))) {
+        return 0;
+    }
+
     for (Memo *m = t->memo; m != NULL; m = m->next) {
         if (m->type == type) {
 #if defined(Py_DEBUG)
diff --git a/Parser/pegen.h b/Parser/pegen.h
index cbf44ba474fa395..12c038fa18a9080 100644
--- a/Parser/pegen.h
+++ b/Parser/pegen.h
@@ -42,6 +42,10 @@ typedef struct {
     int level;
     int lineno, col_offset, end_lineno, end_col_offset;
     Memo *memo;
+    // Filter over the rule types present in `memo` (bit `type & 63` is set
+    // for every entry): lets lookups skip walking the list on definite
+    // misses, which are the common case.
+    uint64_t memo_mask;
     PyObject *metadata;
 } Token;