cubic
Try cubic for free
Updated

GPT-5.5

As of Sep 26, 2026, GPT-5.5 ranks #1 of 9 on Bug Bench with a relative bug rate of 0.47 (53% fewer counted bugs than average).

Relative bug rate
0.47
Rank of 9
#1

1.00 = average · lower is better

Provider
OpenAI
Released
Apr 23, 2026
Price per 1M tokens
$5 in · $30 out
First seen
Week of Jun 1

Nearby in the ranking

Full ranking →
GPT-5.5 and the models ranked closest to it, by relative bug rate.
RankModelRate
1GPT-5.5
0.47
2Claude Fable 5
0.70
3GPT-5.6 Sol
0.73

Is GPT-5.5 getting worse?

All weekly trends →

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

Where it does better or worse

# = rank among models

By coding agent

GPT-5.5 relative bug rate by coding agent.
RankAgentRate
1Codex
0.42
2Cursor
0.88

How widely GPT-5.5 is used

Share of AI-written code reviewed by cubic
Latest weekWeek of Sep 14
0.1%
This week so farSince Sep 21
0.1%
Peak weekWeek of Jun 15
33.3%

Compare GPT-5.5

Questions and answers

Is GPT-5.5 good for coding?
GPT-5.5 ranks #1 of 9 on Bug Bench, with 53% fewer counted bugs than average (0.47) from Jun 1 to Sep 20, 2026.
Is GPT-5.5 getting worse?
Not enough weekly data yet (0 of 6 weeks).
Which coding agent has the lowest bug rate with GPT-5.5?
Codex (0.42), ahead of Cursor (0.88).
Embed GPT-5.5's Bug Bench badge

Markdown

[![GPT-5.5 on Bug Bench](https://www.cubic.dev/bugbench/badge/gpt-5-5)](https://www.cubic.dev/bugbench/models/gpt-5-5)

HTML

<a href="https://www.cubic.dev/bugbench/models/gpt-5-5"><img src="https://www.cubic.dev/bugbench/badge/gpt-5-5" alt="GPT-5.5 on Bug Bench" /></a>