Which AI model writes the most bugs in Python?

As of Sep 30, 2026, Claude Sonnet 5 had the highest relative bug rate in Python code (1.87, 87% more counted bugs than the Python average).

Updated Sep 30, 2026

Best AI model for Python code

As of Sep 30, 2026, GPT-5.6 Sol had the lowest relative bug rate in Python code (0.80, 20% fewer counted bugs than the Python average), statistically tied with Claude Opus 5.5, among 7 ranked models.

Models ranked by relative bug rate in Python code. Lower values are better.
Rank	Model	Rate
1= (statistically tied)	GPT-5.6 Sol	0.80
2= (statistically tied)	Claude Opus 4.8	0.92
3= (statistically tied)	Claude Fable 5.1	0.97
4= (statistically tied)	Claude Opus 5.5	1.00
5= (statistically tied)	GPT-6 Astra	1.35
6= (statistically tied)	Claude Opus 5	1.44
7	Claude Sonnet 5	1.87

1.00 = Python average · lower is better · = statistically tied

Other languages

TypeScript
JavaScript
Go
Rust
SQL
Shell
Ruby
Swift
PHP
Vue
Terraform
Kotlin
C#
Svelte
Java
Dart

Questions and answers

Which AI model writes the most bugs in Python?: As of Sep 30, 2026, Claude Sonnet 5 had the highest relative bug rate in Python code (1.87, 87% more counted bugs than the Python average).