diff --git a/includes/ast/lexer.vp b/includes/ast/lexer.vp
index 2b82bb8..924b020 100644
--- a/includes/ast/lexer.vp
+++ b/includes/ast/lexer.vp
@@ -473,6 +473,8 @@ def _read_string(lx: Lexer | t.CPtr):
 
     # Read string content
     content_buf_start: t.CSizeT = lx.pos
+    # f-string {} 表达式块嵌套深度（PEP 701 同引号嵌套字符串支持用）
+    fstr_depth: t.CInt = 0
     while lx.pos < lx.len:
         ch = lx.src[lx.pos]
         if is_triple:
@@ -498,6 +500,50 @@ def _read_string(lx: Lexer | t.CPtr):
             lx.pos += 1
             lx.col += 1
         else:
+            if is_fstring:
+                # {{ 与 }} 是转义字面量，不参与深度计数
+                if ch == '{' and lx.pos + 1 < lx.len and lx.src[lx.pos + 1] == '{':
+                    lx.pos += 2
+                    lx.col += 2
+                    continue
+                if ch == '}' and lx.pos + 1 < lx.len and lx.src[lx.pos + 1] == '}':
+                    lx.pos += 2
+                    lx.col += 2
+                    continue
+                if ch == '{':
+                    fstr_depth += 1
+                    lx.pos += 1
+                    lx.col += 1
+                    continue
+                if ch == '}':
+                    if fstr_depth > 0:
+                        fstr_depth -= 1
+                    lx.pos += 1
+                    lx.col += 1
+                    continue
+                if fstr_depth > 0 and (ch == '\'' or ch == '"'):
+                    # {} 表达式块内允许嵌套字符串字面量（PEP 701 同引号嵌套）：
+                    # 跳过到匹配引号，块内引号不终止外层 f-string。
+                    nq: t.CChar = ch
+                    lx.pos += 1
+                    lx.col += 1
+                    while lx.pos < lx.len:
+                        c2: t.CChar = lx.src[lx.pos]
+                        if c2 == '\\':
+                            lx.pos += 2
+                            lx.col += 2
+                            continue
+                        if c2 == nq:
+                            lx.pos += 1
+                            lx.col += 1
+                            break
+                        if c2 == '\n':
+                            # 单行字符串不能跨行
+                            lx.error_count += 1
+                            return
+                        lx.pos += 1
+                        lx.col += 1
+                    continue
             if ch == quote:
                 break
             if ch == '\n':
diff --git a/includes/ast/parser.vp b/includes/ast/parser.vp
index 50cd84c..6bbf8dc 100644
--- a/includes/ast/parser.vp
+++ b/includes/ast/parser.vp
@@ -1286,6 +1286,21 @@ def _parse_fstring(ps: Parser | t.CPtr, content: str,
             fmt_pos: t.CSizeT = 0
             while j < content_len and depth > 0:
                 c: t.CChar = content[j]
+                if c == '\'' or c == '"':
+                    # 跳过表达式块内的嵌套字符串字面量（PEP 701 同引号嵌套），
+                    # 避免其内部字符干扰 {} 深度/转换/格式分隔符定位
+                    nq: t.CChar = c
+                    j += 1
+                    while j < content_len:
+                        c2: t.CChar = content[j]
+                        if c2 == '\\':
+                            j += 2
+                            continue
+                        if c2 == nq:
+                            j += 1
+                            break
+                        j += 1
+                    continue
                 if c == '{':
                     depth += 1
                 elif c == '}':
@@ -1307,6 +1322,7 @@ def _parse_fstring(ps: Parser | t.CPtr, content: str,
                         fmt_pos = j
                 j += 1
             if depth != 0:
+                _record_parse_err(ps)
                 ps.error_count += 1
                 break
             # Extract expression text
@@ -1317,6 +1333,7 @@ def _parse_fstring(ps: Parser | t.CPtr, content: str,
                 expr_end = fmt_pos
             expr_len: t.CSizeT = expr_end - i - 1
             if expr_len == 0:
+                _record_parse_err(ps)
                 ps.error_count += 1
                 i = j + 1
                 continue
@@ -1356,6 +1373,7 @@ def _parse_fstring(ps: Parser | t.CPtr, content: str,
                 lit_pos += 1
                 i += 2
             else:
+                _record_parse_err(ps)
                 ps.error_count += 1
                 i += 1
         else: