Updated

GPT-5.6 Luna

As of Sep 26, 2026, GPT-5.6 Luna ranks #5 of 7 on Bug Bench with a relative bug rate of 1.28 (28% more counted bugs than average), statistically tied with GPT-6 Astra and Claude Fable 5.1.

Relative bug rate
1.28
Rank of 7
#5
Tied with
GPT-6 Astra and Claude Fable 5.1

1.00 = average · lower is better

First seen
Week of Jul 6

Nearby in the ranking

GPT-5.6 Luna and the models ranked closest to it, by relative bug rate.
RankModelRate
3= (statistically tied)GPT-6 Astra
1.16
4= (statistically tied)Claude Fable 5.1
1.17
5GPT-5.6 Luna
1.28
6Claude Opus 5
1.58
7Claude Sonnet 5
2.10

Is GPT-5.6 Luna getting worse?

Not enough weekly data yet (0 of 6 weeks).

Where it does better or worse

# = rank among models

By coding agent

GPT-5.6 Luna relative bug rate by coding agent.
RankAgentRate
2Codex
1.32

How widely GPT-5.6 Luna is used

Share of AI-written code reviewed by cubic
Latest weekWeek of Sep 14
1.7%
This week so farSince Sep 21
0.9%
Peak weekWeek of Aug 24
2.8%

Compare GPT-5.6 Luna

Questions and answers

Is GPT-5.6 Luna good for coding?
GPT-5.6 Luna ranks #5 of 7 on Bug Bench, with 28% more counted bugs than average (1.28) from Jun 1 to Sep 20, 2026.
Is GPT-5.6 Luna getting worse?
Not enough weekly data yet (0 of 6 weeks).
Which coding agent has the lowest bug rate with GPT-5.6 Luna?
Codex, at 1.32.
Embed GPT-5.6 Luna's Bug Bench badge

Markdown

[![GPT-5.6 Luna on Bug Bench](https://www.cubic.dev/bugbench/badge/gpt-5-6-luna)](https://www.cubic.dev/bugbench/models/gpt-5-6-luna)

HTML

<a href="https://www.cubic.dev/bugbench/models/gpt-5-6-luna"><img src="https://www.cubic.dev/bugbench/badge/gpt-5-6-luna" alt="GPT-5.6 Luna on Bug Bench" /></a>