AI reference score 6.83 ± 0.95 · max (best) Last 4 weekly runs · 12 images
Tier comparison
TierSamplesAI reference scoreValid SVG rateMean timeSamples
max
GPT-6 Luna · max · 2026-10-26 · #1 · 8.11
6.83 ± 0.95Last 4 weekly runs · 12 images 100%45 s3
medium default
GPT-6 Luna · medium · 2026-10-26 · #3 · 7.06
5.64 ± 0.84Last 4 weekly runs · 12 images 100%32 s3
low
GPT-6 Luna · low · 2026-10-26 · #3 · 7.26
5.60 ± 1.28Last 4 weekly runs · 12 images 100%27 s3

Latest samples

max

AI reference score 6.83 ± 0.95 Last 4 weekly runs · 12 images Normal

medium

AI reference score 5.64 ± 0.84 Last 4 weekly runs · 12 images Normal

low

AI reference score 5.60 ± 1.28 Last 4 weekly runs · 12 images Normal

Score trend

max Normal

2026-10-08 · 7.302026-10-09 · 7.832026-10-10 · 6.702026-10-11 · 7.822026-10-12 · 7.342026-10-13 · 6.132026-10-14 · 7.112026-10-15 · 7.412026-10-16 · 6.422026-10-17 · 7.962026-10-18 · 7.382026-10-19 · 7.492026-10-20 · 6.962026-10-21 · 7.152026-10-22 · 6.512026-10-23 · 7.482026-10-24 · 6.452026-10-25 · 6.902026-10-26 · 6.47

19 records in the last 30 days

medium Normal

2026-10-08 · 5.302026-10-09 · 5.322026-10-10 · 6.242026-10-11 · 5.272026-10-12 · 5.772026-10-13 · 6.872026-10-14 · 6.592026-10-15 · 6.652026-10-16 · 5.982026-10-17 · 6.072026-10-18 · 6.542026-10-19 · 5.312026-10-20 · 6.102026-10-21 · 7.032026-10-22 · 7.652026-10-23 · 5.712026-10-24 · 5.822026-10-25 · 5.362026-10-26 · 5.67

19 records in the last 30 days

low Normal

2026-10-08 · 6.382026-10-09 · 5.352026-10-10 · 5.622026-10-11 · 6.232026-10-12 · 4.882026-10-13 · 6.292026-10-14 · 6.002026-10-15 · 5.282026-10-16 · 5.262026-10-17 · 5.012026-10-18 · 5.952026-10-19 · 5.422026-10-20 · 6.252026-10-21 · 6.252026-10-22 · 6.102026-10-23 · 6.502026-10-24 · 5.622026-10-25 · 4.282026-10-26 · 6.01

19 records in the last 30 days

Earlier samples

max

medium

low

Versions by score

  1. GPT-6 Luna
    GPT-6 Luna · max
    max 6.83Last 4 weekly runs · 12 images
  2. GPT-5.6 Luna
    GPT-5.6 Luna · max
    max 5.92Last 4 weekly runs · 12 images

Scores by criterion

Scores by tier
Scoring criteria max medium low
Pelican likeness6.426.425.75
Bicycle correctness7.336.256.08
Riding pose5.423.755.25
Composition6.926.837.17
Obvious errors (higher is better)6.084.836.17

Release notes

Original (Chinese)与 GPT-6 Sol 同日发布。当天在 ChatGPT Work、Codex 和 API(gpt-6-luna)可用;Free/Go 可在桌面端使用。公告写明 Chat 暂未上线。2026-10-08 起 GPT-6 Luna 向 Free/Go 的 Chat 推出。

· View release source ↗