Switch to Kodezi Code
Kodezi CLILanguage Support
Chronos' debugging performance across different programming languages.
Chronos-1 maintains consistently strong performance across multiple programming languages, each evaluated on real-world bugs.
Language-wise Accuracy
- Python: 68.7% ± 2.1%
- JavaScript: 64.2% ± 2.3%
- Java: 63.9% ± 2.2%
- Go: 66.8% ± 2.4%
- C++: 61.2% ± 2.6%
- Rust: 59.8% ± 2.7%
General-purpose models achieve 3.7–15.8% on the same datasets.
Chronos’ debugging-first training corpus enables deep understanding of:
- language-specific idioms
- framework patterns
- runtime behavior differences
- ecosystem-level failure modes