diff --git a/assayer/judge.py b/assayer/judge.py
index c141cc1..d7c8162 100644
--- a/assayer/judge.py
+++ b/assayer/judge.py
@@ -2,9 +2,10 @@
 
 import json
 import re
-import sys
 from dataclasses import dataclass, field
 
+import click
+
 from assayer.models import ModelResult
 from assayer.runner import _make_provider
 
@@ -72,7 +73,7 @@ async def run_judge(
 ) -> JudgeResult | None:
     valid = [r for r in results if not r.error and r.output]
     if len(valid) < 2:
-        print("Judge skipped: fewer than 2 successful results.", file=sys.stderr)
+        click.echo("Judge skipped: fewer than 2 successful results.", err=True)
         return None
 
     judge_prompt = _build_prompt(prompt, valid, criteria)
@@ -81,15 +82,15 @@ async def run_judge(
     try:
         result = await provider.run(judge_prompt)
     except Exception as exc:
-        print(f"Judge call failed: {exc}", file=sys.stderr)
+        click.echo(f"Judge call failed: {exc}", err=True)
         return None
 
     if result.error:
-        print(f"Judge call failed: {result.error}", file=sys.stderr)
+        click.echo(f"Judge call failed: {result.error}", err=True)
         return None
 
     try:
         return _parse_response(result.output)
     except Exception as exc:
-        print(f"Judge response could not be parsed: {exc}", file=sys.stderr)
+        click.echo(f"Judge response could not be parsed: {exc}", err=True)
         return None
diff --git a/assayer/renderer.py b/assayer/renderer.py
index 90a3755..e12f900 100644
--- a/assayer/renderer.py
+++ b/assayer/renderer.py
@@ -7,6 +7,7 @@
 
 from assayer.judge import JudgeResult
 from assayer.models import ModelResult
+from assayer.scorer import readability_stats
 
 console = Console()
 
@@ -51,6 +52,7 @@ def render_run(
 
     if similarity is not None:
         render_similarity_matrix(results, similarity)
+        render_readability(results)
 
     if judge_result is not None:
         render_judge(judge_result)
@@ -83,6 +85,29 @@ def render_similarity_matrix(
     console.print(table)
 
 
+def render_readability(results: list[ModelResult]) -> None:
+    valid = [r for r in results if not r.error and r.output]
+    if not valid:
+        return
+
+    table = Table(title="Readability", show_header=True)
+    table.add_column("Model", style="bold")
+    table.add_column("Words", justify="right")
+    table.add_column("Sentences", justify="right")
+    table.add_column("Avg sentence length", justify="right")
+
+    for r in valid:
+        stats = readability_stats(r.output)
+        table.add_row(
+            r.model,
+            str(int(stats["word_count"])),
+            str(int(stats["sentence_count"])),
+            f"{stats['avg_sentence_length']:.1f}",
+        )
+
+    console.print(table)
+
+
 def render_judge(judge_result: JudgeResult) -> None:
     console.print(Rule("[bold]Judge[/bold]"))
     console.print(f"[bold green]Winner:[/bold green] {judge_result.winner}")