diff --git a/.github/workflows/performance.yml b/.github/workflows/performance.yml
index 916f865d..2987224a 100644
--- a/.github/workflows/performance.yml
+++ b/.github/workflows/performance.yml
@@ -29,15 +29,25 @@ jobs:
         id: bench
         run: |
           echo "Running performance benchmarks..."
-          pnpm run bench 2>&1 | tee bench-output.txt
+          pnpm exec vitest bench --run --config vitest.bench.config.mjs --outputJson bench-results.json 2>&1 | tee bench-output.txt
 
       - name: Generate benchmark summary
         if: always()
         run: |
-          echo "## 📊 Performance Benchmark Results" >> $GITHUB_STEP_SUMMARY
-          echo "" >> $GITHUB_STEP_SUMMARY
+          # Parse JSON and create markdown table
+          if [ -f bench-results.json ]; then
+            node scripts/format-benchmarks.mjs bench-results.json >> $GITHUB_STEP_SUMMARY
+          else
+            echo "## 📊 Performance Benchmark Results" >> $GITHUB_STEP_SUMMARY
+            echo "" >> $GITHUB_STEP_SUMMARY
+            echo "⚠️ No benchmark results found" >> $GITHUB_STEP_SUMMARY
+            echo "" >> $GITHUB_STEP_SUMMARY
+          fi
+
+          # Show terminal output
+          echo "### Terminal Output" >> $GITHUB_STEP_SUMMARY
           echo '```' >> $GITHUB_STEP_SUMMARY
-          cat bench-output.txt >> $GITHUB_STEP_SUMMARY
+          sed 's/\x1b\[[0-9;]*m//g' bench-output.txt >> $GITHUB_STEP_SUMMARY
           echo '```' >> $GITHUB_STEP_SUMMARY
 
       - name: Check for regressions (informational)
diff --git a/scripts/format-benchmarks.mjs b/scripts/format-benchmarks.mjs
new file mode 100755
index 00000000..48645374
--- /dev/null
+++ b/scripts/format-benchmarks.mjs
@@ -0,0 +1,65 @@
+#!/usr/bin/env node
+
+import { readFileSync } from 'fs';
+
+function addCommas(num) {
+  const str = num.toString();
+  if (str.length <= 3) return str;
+  return addCommas(str.slice(0, -3)) + ',' + str.slice(-3);
+}
+
+function formatNumber(num, decimals = 4) {
+  const multiplier = Math.pow(10, decimals);
+  return (Math.round(num * multiplier) / multiplier).toString();
+}
+
+function formatBenchmarkResults(jsonFile) {
+  const data = JSON.parse(readFileSync(jsonFile, 'utf8'));
+
+  const lines = [];
+  lines.push('## 📊 Performance Benchmark Results');
+  lines.push('');
+  lines.push('| Benchmark | Hz | Min | Max | Mean | P75 | P99 | P995 | P999 |');
+  lines.push('|-----------|-------|------|------|------|------|------|------|------|');
+
+  let totalBenchmarks = 0;
+
+  for (const file of data.files) {
+    for (const group of file.groups) {
+      for (const benchmark of group.benchmarks) {
+        totalBenchmarks++;
+
+        const row = [
+          benchmark.name,
+          addCommas(Math.floor(benchmark.hz)),
+          formatNumber(benchmark.min),
+          formatNumber(benchmark.max),
+          formatNumber(benchmark.mean),
+          formatNumber(benchmark.p75),
+          formatNumber(benchmark.p99),
+          formatNumber(benchmark.p995),
+          formatNumber(benchmark.p999)
+        ];
+
+        lines.push('| ' + row.join(' | ') + ' |');
+      }
+    }
+  }
+
+  lines.push('');
+  lines.push(`_Total benchmarks: ${totalBenchmarks}_`);
+  lines.push('');
+
+  return lines.join('\n');
+}
+
+// Main execution
+const jsonFile = process.argv[2] || 'bench-results.json';
+
+try {
+  const markdown = formatBenchmarkResults(jsonFile);
+  console.log(markdown);
+} catch (error) {
+  console.error('Error formatting benchmark results:', error.message);
+  process.exit(1);
+}