Updated
Best AI coding models, October 2026 edition
As of Oct 31, 2026, GPT-6 Luna had the lowest relative bug rate on Bug Bench (0.34, 66% fewer counted bugs than average), among 7 ranked AI coding models. Final October 2026 ranking.
| Rank | Model | Rate |
|---|---|---|
| 1 | GPT-6 Luna | 0.34 |
| 2= (statistically tied) | GPT-6.1 Sol | 0.86 |
| 3= (statistically tied) | Claude Sonnet 5.5 | 0.88 |
| 4= (statistically tied) | Claude Fable 5.1 | 1.02 |
| 5= (statistically tied) | GPT-6 Astra | 1.04 |
| 6= (statistically tied) | Claude Opus 5.5 | 1.25 |
| 7= (statistically tied) | Claude Opus 5 | 1.26 |
1.00 = average · lower is better · = statistically tied
Most-used models and coding agents
Share of AI-written code, week of Oct 19
Models
- Claude Opus 5.5
- 59.7%
- GPT-6.1 Sol
- 12.5%
- GPT-6 Astra
- 5.9%
- Claude Sonnet 5.5
- 4.5%
- Claude Fable 5.1
- 4.1%
Coding agents
- Claude Code
- 76.6%
- Codex
- 20.3%
- Cursor
- 1.3%