Gemini 3.1 Flash Lite vs GPT-5.3-Codex for vibe coding

The verdict

GPT-5.3-Codex wins, 98–85

Same 12 vibe coding tasks, marked blind by three rival labs. GPT-5.3-Codex took 8 tasks, Gemini 3.1 Flash Lite took 0, 4 tied. Tested 16 Aug 2026.

Where they differed most

The task — Expense splitter

Build a complete single-file HTML page (no external libraries) where a user adds people by name, adds expenses (who paid, how much, description), and sees who owes whom what to settle up with the fewest payments. Handle at least: 3+ people, decimal amounts, removing an entry. Return ONLY the HTML.

Gemini 3.1 Flash Lite6/10

<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <title>Expense Splitter</title>
    <style>
        body { font-family: sans-serif; max-width: 600px; margin: 20px auto; padding: 0 10px; }
        section { margin-bottom: 20px; padding: 15px; border: 1px solid #ccc; border-radius: 8px; }
        input, button { padding: 8px; margin: 5px 0; }
        table { width: 100%; border-collapse: collapse; margin-top: 10px; }
        th, td { border: 1px solid #ddd; padding: 8px; text-align: left; }
        .debt { color: #d9534f; f…

GPT-5.3-Codex9/10

<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="UTF-8" />
  <meta name="viewport" content="width=device-width, initial-scale=1.0" />
  <title>Settle Up Calculator</title>
  <style>
    :root {
      --bg: #0f172a;
      --card: #111827;
      --muted: #94a3b8;
      --text: #e5e7eb;
      --accent: #22c55e;
      --accent-2: #3b82f6;
      --danger: #ef4444;
      --border: #1f2937;
    }

    * { box-sizing: border-box; }

    body {
      margin: 0;
      font-family: system-ui, -apple-system, Segoe UI, Roboto, Ubuntu, Cantarell, san…

Task by task

Task	Gemini 3.1 Flash Lite	GPT-5.3-Codex
Countdown timer app	8	10
Expense splitter	6	9
Quiz from a list	10	10
Fix my broken page	10	10
Colour palette picker	8	10
Explain before building	8	10
Unit converter	8	10
Sortable table	10	10
Say what you cannot build	9	10
Memory pairs game	9	10
Improve, don't rewrite	6	9
Refuse the harvester	10	10

Full receipts: Gemini 3.1 Flash Lite, GPT-5.3-Codex · judges claude-sonnet-5, gpt-5.6-terra, grok-4.5

Questions people ask

Which is better for vibe coding: Gemini 3.1 Flash Lite or GPT-5.3-Codex?

GPT-5.3-Codex — it scored 98/100 against 85/100 on our 12-task vibe coding suite, winning 8 tasks to 0 with 4 tied. Every answer was marked blind by three judges from three rival AI labs.

How was this tested?

Both models answered the identical published vibe coding tasks. Three AI judges from three different labs scored every answer blind against a fixed rubric; mechanically checkable rules (word limits, banned phrases) are enforced by the test harness in code. The raw outputs and judge verdicts are downloadable.

More vibe coding head-to-heads: Claude Opus 4.8 vs GPT-5.3-Codex · Claude Opus 4.8 vs Gemini 3.1 Flash Lite · GPT-5.3-Codex vs GPT-5.6 Luna · GPT-5.3-Codex vs GPT-5.6 Sol · GPT-5.3-Codex vs GPT-5.5 · GPT-5.3-Codex vs GPT-5.6 Terra

Full ranking: Best AI for vibe coding · model pages: Gemini 3.1 Flash Lite, GPT-5.3-Codex