AI Model Rankings
Rankings based on real usage data from developers using Source Trace.
Total Lines Tracked
...
Avg AI Survival Rate
...
Total Commits
...
AI Agents / Models
... / ...
Daily Activity
AI generates a lot of code, some is one-shot, some is rewritten before commit, some removed later. Get the true data about code churn rate.
Lines by Model
Lines of code added per model. Find which AI is most used by real developers.
Survival Rate
AI generated code that survives. The higher the better - 100% means the model is good at one-shot.
Top Models
See which AI models are most used by the community.
| Model | Lines | Commits | Survival Rate |
|---|---|---|---|
| Claude Sonnet 3.5 | 890,234 | 3,421 | 78.5% |
| GPT-4 | 567,890 | 2,156 | 71.2% |
| Claude Opus | 456,123 | 1,843 | 82.1% |
| Gemini Pro | 234,567 | 987 | 65.8% |
| GPT-3.5 Turbo | 156,789 | 654 | 59.3% |
Last updated: 21/11/2025, 12:00:00
All data is aggregated and anonymized. No individual repositories or users are identified.