diff --git a/scripts/run_tests_batched.py b/scripts/run_tests_batched.py
index bbfa9027..6139665d 100644
--- a/scripts/run_tests_batched.py
+++ b/scripts/run_tests_batched.py
@@ -1,34 +1,84 @@
 import argparse
+import json
+import subprocess
 import sys
+import time
 from pathlib import Path
 
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent / "tests"))
 
 from categorizer import categorize_all
-from batcher import plan
+from batcher import plan, Batch
 
-def _print_plan(records, options) -> int:
- batches = plan(records, include_opt_in=options.include_opt_in, xdist=not options.no_xdist)
- for b in batches:
-  status = "SKIP" if b.skip_reason else "RUN"
-  print(f"[{status}] {b.label}: {len(b.files)} files, est {b.estimated_seconds:.1f}s, args={b.pytest_args}")
+def _parse_tiers(s: str) -> set[str]:
+ return {t.strip() for t in s.split(",") if t.strip()}
+
+def _durations_path(tests_dir: Path) -> Path:
+ return tests_dir / ".test_durations.json"
+
+def _load_durations(p: Path) -> dict[str, float]:
+ if not p.exists():
+  return {}
+ try:
+  with p.open("r", encoding="utf-8") as f:
+   return json.load(f)
+ except (json.JSONDecodeError, OSError):
+  return {}
+
+def _save_durations(p: Path, durations: dict[str, float]) -> None:
+ tmp = p.with_suffix(".json.tmp")
+ with tmp.open("w", encoding="utf-8") as f:
+  json.dump(durations, f, indent=2, sort_keys=True)
+ tmp.replace(p)
+
+def _parse_durations_from_pytest_output(stdout: str) -> dict[str, float]:
+ out: dict[str, float] = {}
+ for line in stdout.splitlines():
+  line = line.strip()
+  if "::" not in line or " " not in line:
+   continue
+  parts = line.rsplit(None, 1)
+  if len(parts) != 2:
+   continue
+  nodeid, time_str = parts
+  try:
+    out[nodeid] = float(time_str.rstrip("s"))
+  except ValueError:
+   continue
+ return out
+
+def _run_batch(b: Batch, durations: dict[str, float]) -> tuple[int, float, dict[str, float]]:
+ if b.skip_reason:
+  return 0, 0.0, {}
+ cmd = ["uv", "run", "pytest", "-v", "--durations=0"] + b.pytest_args + [str(f) for f in b.files]
+ print(f"\n>>> Running {b.label} ({len(b.files)} files)")
+ t0 = time.monotonic()
+ proc = subprocess.run(cmd, capture_output=True, text=True)
+ elapsed = time.monotonic() - t0
+ new_durs = _parse_durations_from_pytest_output(proc.stdout)
+ tail = proc.stdout[-2000:] if proc.returncode != 0 else f"<<< {b.label} PASS in {elapsed:.1f}s"
+ print(tail)
+ if proc.returncode != 0:
+  print(f"<<< {b.label} FAIL (exit {proc.returncode}) in {elapsed:.1f}s")
+  print(proc.stderr[-1000:])
+ return proc.returncode, elapsed, new_durs
+
+def _print_summary(results: list[tuple[Batch, int, float]]) -> int:
+ print("\n" + "=" * 60)
+ print("SUMMARY")
+ print("=" * 60)
+ worst = 0
+ for b, code, elapsed in results:
   if b.skip_reason:
-   print(f"    reason: {b.skip_reason}")
- return 0
-
-def _print_audit(records, strict: bool) -> int:
- auto = [r for r in records if r.source == "auto"]
- print(f"Auto-inferred (unclassified) records: {len(auto)}")
- for r in auto:
-  print(f" {r.filename}: fc={r.fixture_class.value}, subs={r.subsystems}, bg={r.batch_group}")
- if strict:
-  bad = [r for r in auto if len(r.subsystems) > 1]
-  if bad:
-   print(f"STRICT: {len(bad)} auto-inferred files have multiple subsystems (probably cross-cutting):")
-   for r in bad:
-    print(f" {r.filename}: subs={r.subsystems}")
-   return 1
- return 0
+   status = "SKIPPED"
+  elif code == 0:
+   status = "PASS"
+  else:
+   status = "FAIL"
+  worst = max(worst, code)
+  n = len(b.files)
+  print(f"[{b.tier}] {b.label:40s} {status:8s} {n} files {elapsed:6.1f}s")
+ return worst
 
 def main() -> int:
  p = argparse.ArgumentParser()
@@ -40,14 +90,41 @@ def main() -> int:
  p.add_argument("--plan", action="store_true")
  p.add_argument("--audit", action="store_true")
  p.add_argument("--strict", action="store_true")
+ p.add_argument("--durations", action="store_true", help="Record per-test durations to .test_durations.json")
  options = p.parse_args()
- records = categorize_all(Path(options.tests_dir), Path(options.registry))
+ tiers = _parse_tiers(options.tiers)
+ tests_dir = Path(options.tests_dir)
+ durations_path = _durations_path(tests_dir)
+ durations = _load_durations(durations_path)
+ records = categorize_all(tests_dir, Path(options.registry))
  if options.audit:
-  return _print_audit(records, strict=options.strict)
+  auto = [r for r in records if r.source == "auto"]
+  print(f"Auto-inferred (unclassified) records: {len(auto)}")
+  for r in auto:
+   print(f" {r.filename}: fc={r.fixture_class.value}, subs={r.subsystems}, bg={r.batch_group}")
+  if options.strict:
+   bad = [r for r in auto if len(r.subsystems) > 1]
+  if bad:
+   print(f"STRICT: {len(bad)} auto-inferred files have multiple subsystems:")
+  for r in bad:
+   print(f" {r.filename}: subs={r.subsystems}")
+  return 1
+  return 0
+ batches = plan(records, tiers=tiers, include_opt_in=options.include_opt_in, xdist=not options.no_xdist)
  if options.plan:
-  return _print_plan(records, options)
- print("Phase 1 stub: no actual test execution yet. Use --plan or --audit.")
- return 0
+  for b in batches:
+   status = "SKIP" if b.skip_reason else "RUN"
+  print(f"[{status}] {b.label}: {len(b.files)} files, est {b.estimated_seconds:.1f}s")
+  return 0
+ results: list[tuple[Batch, int, float]] = []
+ merged_durations = dict(durations)
+ for b in batches:
+  code, elapsed, new_durs = _run_batch(b, merged_durations)
+  results.append((b, code, elapsed))
+  merged_durations.update(new_durs)
+ if options.durations:
+  _save_durations(durations_path, merged_durations)
+ return _print_summary(results)
 
 if __name__ == "__main__":
  sys.exit(main())