SEC-Bench Pro: Results as Published

SEC-bench Pro tests whether AI coding agents can find real, previously disclosed security bugs in large, critical software, mainly the V8 and SpiderMonkey JavaScript engines. It contains 183 validated cases, each packaged in a container with a fixed version and a checkable crash result. The agents work without hints such as sanitizer output that names the vulnerable function, and the authors use an LLM judge to grade results.

Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.