Detection rate alone ranks the worst LLM code reviewers first

Sunny Chokshi, a PhD student at the University of the Cumberlands, benchmarked 16 open-weight LLM code reviewers on 180 Python snippets across nine cryptographic misuse classes, running 8,640 scored calls with paired vulnerable and secure twins. Ranking by true positive rate alone put codellama 7B, devstral 24B and qwen2.5-coder 0.5B near the top, but those models flagged most or all secure twins, while gpt-oss 20B had the highest balanced accuracy at 0.952 with a 93.7% TPR and 3.3% FPR. Chokshi concludes that detection rate should be reported alongside false positive rate using matched secure twins, and notes the preprint is not peer reviewed and covers only open models and Python.

When I tested 16 open LLM code reviewers, the first table in my paper showed which of them separate secure code from vulnerable code at all. It is worth reading on its own. I used 180 Python snippets across nine cryptographic misuse classes, ten pairs per class. Each pair has a vulnerable snippet and a secure twin that does the same task correctly. I ran 16 open models, from 0.5B to 120B parameters, with three repeats per snippet, which gives 8,640 scored calls. For each model I report the true positive rate TPR, vulnerable snippets flagged , the false positive rate FPR, secure twins flagged , and balanced accuracy, the average of TPR and one minus FPR. A model that flags everything scores 0.5. Judged on TPR alone, codellama 7B, devstral 24B, and qwen2.5-coder 0.5B would rank near the top, each at or near 100%. Their FPRs tell a different story. codellama 7B and qwen2.5-coder 0.5B flagged 539 and 540 of 540 scored calls. devstral 24B flagged 82.6% of the secure twins. Three more models codellama 34B, qwen2.5-coder 1.5B and 3B also flagged more than 70% of secure code. The 0.5B model also left out the requested verdict line in all but 37 of its 540 replies. With the secure twins in the table, the order changes. Of the 16 models, gpt-oss 20B had the highest balanced accuracy, 0.952, with a TPR of 93.7% and an FPR of 3.3%. Ranked by TPR alone, these models would sit at or near the top. Size and score go together to some degree, but size does not settle the ranking. Over all 16 models, parameter count and balanced accuracy are related Spearman 0.76 . Across families it did not order the results: codellama 34B 0.609 and devstral 24B 0.587 were among the weakest of the larger models. Even within gpt-oss, the 120B model 0.931 scored below the 20B 0.952 . What I take from this test: a detection rate on its own rewarded models that said "vulnerable" almost every time. I would report the false positive rate next to it and use secure twins that match the vulnerable code. The snippets are short Python functions, reviewed one at a time, so the result speaks to a single-snippet review pass. The models are open models run with temperature 0. I tested only open-weight models and only Python, so I make no claim about closed models or other languages. TPR counts a vulnerable verdict given for any reason, and I did not check whether each detection named the right misuse. This is a preprint, it is not peer reviewed, and it carries the notice that it has been submitted to the IEEE for possible publication. The same paper goes on to test what happens when the same misuse is written indirectly. I covered one result from that part in my earlier post on a hex key size https://sunny-chokshi.github.io/blog/hex-key-size-false-alarms.html . Preprint on Zenodo doi:10.5281/zenodo.23351769 https://doi.org/10.5281/zenodo.23351769 · Paper page and PDF https://sunny-chokshi.github.io/papers/does-indirection-hurt.html Related work in the same line, an earlier benchmark: CryptoBench doi:10.5281/zenodo.23067051 https://doi.org/10.5281/zenodo.23067051 . Cite as: Chokshi, S. 2026 . Does Indirection Hurt? Paired Evidence on LLM Review of Indirectly Written Cryptographic API Misuse Preprint . Zenodo. https://doi.org/10.5281/zenodo.23351769 https://doi.org/10.5281/zenodo.23351769 Dataset: Chokshi, S. 2026 . CryptoBench: A benchmark of cryptographic API misuse for evaluating LLM code reviewers Dataset . Zenodo. https://doi.org/10.5281/zenodo.23067051 https://doi.org/10.5281/zenodo.23067051 Sunny Chokshi, PhD student, Information Technology, University of the Cumberlands. Research: adversarial robustness of LLMs in cybersecurity. ORCID https://orcid.org/0009-0003-4738-7759 · GitHub https://github.com/sunny-chokshi Originally published at https://sunny-chokshi.github.io/blog/reviewer-that-flags-everything.html https://sunny-chokshi.github.io/blog/reviewer-that-flags-everything.html