XRepoTest - Julia - Line Coverage: leaderboard
XRepoTest (arXiv 2608.25939v1, Appendix Table A.4): repository-level unit-test generation in real open-source Julia projects, standard single-turn setting with greedy decoding, with generated tests executed in containers. Focal-function line coverage of the generated tests, normalized over all focal functions, percent. Higher is better.
Metric: Focal line coverage (%; of all focal functions). Source: arxiv.org. Saturation forecast: Not forecast. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 | 31.92 | #115 |
| 2 | Claude Sonnet 4.5 | 31.58 | #148 |
| 3 | DeepSeek V4 Pro | 26.83 | #112 |
| 4 | Llama 3.3 70B Instruct | 25.42 | #541 |
| 5 | GPT-OSS-120B | 24.27 | #346 |
| 6 | Claude Haiku 4.5 (20251001) | 23.96 | #264 |
| 7 | DeepSeek V3.2 | 21.94 | #210 |
| 8 | Codestral-22B-v0.1 | 21.17 | #850 |
| 9 | Yi-Coder-9B-Chat | 19.16 | #1244 |
| 10 | MiniMax-M2.7 | 18.12 | #270 |
| 11 | Qwen3 Coder Next | 17.38 | #331 |
| 12 | GLM-5 | 13.54 | #149 |
| 13 | Kimi K2.5 | 8.7 | #152 |
| 14 | Qwen 3 8B | 0.97 | #678 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=xrepotest-julia-line-coverage · How It Works · Data refreshed daily, snapshot 2026-11-05.