vibecodingnav
中/EN
Models/gpt-6-luna

GPT 6 Luna

Seventh on the Android long board at 10.7%, run cost $5.3. Cheap, and at the bottom of the listed pass rates.

OpenAI · Proprietary · Fetched 2026-10-10

LMArena WebDev
1581
#34 / 142 · ±7
LMArena Image-to-WebDev
1570
#25 / 64 · ±12
Android Bench
10.7%
#7 / 10 · 4.0–17.3

Benchmark scores

Boards older than 90 days are not counted as recommendations
BenchmarkSettingScoreRankCostAs of
LMArena WebDevmax1581±7#34—2026-10-08
LMArena Image-to-WebDevmax1570±12#25—2026-10-09
Android Benchcodex10.74.0–17.3#7$5.32026-10-06
Terminal-Bench 4.0max · Codex16.4±2.7#31$72.142026-10-06
Supabase Evalsmedium · Codex63.8#6—2026-10-10
Next.js Agent Evalshigh · Codex77.0#13$0.0112026-09-25
Convex Evals1 run67.86#8$0.082026-10-08
SWE-bench Verified—This board has stopped updating——2026-02-26
Aider Polyglot—This board has stopped updating——2025-11-20

Use it or not

Why
  • A $5.3 long-horizon run, the lowest cost on this board.
When not
  • 10.7% pass rate. Do not leave multi-module work to it alone.

Appears in these scenarios

Collected from scenario boards
Build a Web frontend#13Next.js 77.0%Build an Android app#7Android Bench 10.7%Write CLI, scripts, and DevOps#31Terminal-Bench 16.4%