diff --git a/packages/x-markdown/src/plugins/Latex/__tests__/index.test.tsx b/packages/x-markdown/src/plugins/Latex/__tests__/index.test.tsx
index f2d686429b..ef39b0f030 100644
--- a/packages/x-markdown/src/plugins/Latex/__tests__/index.test.tsx
+++ b/packages/x-markdown/src/plugins/Latex/__tests__/index.test.tsx
@@ -17,6 +17,43 @@ describe('LaTeX Plugin', () => {
     expect(container).toMatchSnapshot();
   });
 
+  it.each([
+    'Downgrade Max ($100) to Pro ($20)',
+    '把 Max 5x（$100）降级到 Pro（$20）',
+    'The items cost $10, $20, and $30, respectively.',
+  ])('should not interpret currency amounts as LaTeX: %s', (content) => {
+    const warnSpy = jest.spyOn(console, 'warn').mockImplementation(() => {});
+    const { container } = render(
+      <XMarkdown config={{ extensions: latexPlugin() }}>{`**${content}**`}</XMarkdown>,
+    );
+
+    expect(container.querySelector('.katex')).not.toBeInTheDocument();
+    expect(container.querySelector('strong')).toHaveTextContent(content);
+    expect(warnSpy.mock.calls.flat().join(' ')).not.toContain('unicodeTextInMathMode');
+    warnSpy.mockRestore();
+  });
+
+  it('should still render formulas that start with a number', () => {
+    const { container } = render(
+      <XMarkdown config={{ extensions: latexPlugin() }}>{'$2x + 1$'}</XMarkdown>,
+    );
+
+    expect(container.querySelector('.katex')).toBeInTheDocument();
+  });
+
+  it.each([
+    '$ x$',
+    '$x $',
+    '$x$2',
+  ])('should follow single-dollar delimiter rules: %s', (content) => {
+    const { container } = render(
+      <XMarkdown config={{ extensions: latexPlugin() }}>{content}</XMarkdown>,
+    );
+
+    expect(container.querySelector('.katex')).not.toBeInTheDocument();
+    expect(container).toHaveTextContent(content);
+  });
+
   it('should render inline LaTeX with $$\n..\n$$ syntax', () => {
     const { container } = render(
       <XMarkdown config={{ extensions: latexPlugin() }}>
diff --git a/packages/x-markdown/src/plugins/Latex/index.ts b/packages/x-markdown/src/plugins/Latex/index.ts
index 33cd2fe9f6..d978a73dbb 100644
--- a/packages/x-markdown/src/plugins/Latex/index.ts
+++ b/packages/x-markdown/src/plugins/Latex/index.ts
@@ -3,8 +3,8 @@ import type { TokenizerAndRendererExtension } from 'marked';
 
 import 'katex/dist/katex.min.css';
 
-const inlineRuleNonStandard =
-  /^(?:\${1,2}([^$]{1,10000}?)\${1,2}|\\\(([\s\S]{1,10000}?)\\\)|\\\[((?:\\.|[^\\]){1,10000}?)\\\])/;
+const inlineDollarRule = /^(\${1,2})([^$]{1,10000}?)\1/;
+const inlineRuleNonStandard = /^(?:\\\(([\s\S]{1,10000}?)\\\)|\\\[((?:\\.|[^\\]){1,10000}?)\\\])/;
 const blockRule =
   /^(\${1,2})\n([\s\S]{1,10000}?)\n\1(?:\s*(?:\n|$))|^\\\[((?:\\.|[^\\]){1,10000}?)\\\]/;
 
@@ -28,6 +28,14 @@ function replaceAlign(text: string) {
   return text ? text.replace(/\{align\*\}/g, '{aligned}') : text;
 }
 
+// Follow Pandoc's single-dollar delimiter rules to avoid parsing currency as math.
+function isValidInlineDollarMatch(match: RegExpMatchArray, src: string) {
+  const [, delimiter, text] = match;
+  if (delimiter === '$$') return true;
+
+  return !/^\s|\s$/.test(text) && !/^\d/.test(src.slice(match[0].length));
+}
+
 function createRenderer(options: KatexOptions, newlineAfter: boolean) {
   return (token: Token) =>
     katex.renderToString(token.text, {
@@ -49,15 +57,19 @@ function inlineKatex(renderer: Render, replaceAlignStart: boolean) {
       return indices.length > 0 ? Math.min(...indices) : undefined;
     },
     tokenizer(src: string) {
-      const match = src.match(inlineRuleNonStandard);
+      const dollarMatch = src.match(inlineDollarRule);
+      if (dollarMatch && !isValidInlineDollarMatch(dollarMatch, src)) return;
+
+      const nonStandardMatch = src.match(inlineRuleNonStandard);
+      const match = dollarMatch || nonStandardMatch;
       if (!match) return;
 
-      const rawText = match[1] || match[2] || match[3] || '';
+      const rawText = dollarMatch?.[2] || nonStandardMatch?.[1] || nonStandardMatch?.[2] || '';
       const text = replaceAlignStart ? replaceAlign(rawText.trim()) : rawText.trim();
 
       // 对于 \[...\] 语法，如果内容包含换行，标记为块级公式
       // 注意：换行检测必须在 trim 之前进行
-      const isBracketSyntax = match[3] !== undefined;
+      const isBracketSyntax = nonStandardMatch?.[2] !== undefined;
       const hasNewline = rawText.includes('\n');
 
       return {