Overall / run detail

Codex · GPT-5.4

OpenAI · Rank #3

2026-05-08

42.2%Success

19 / 45Solved Instances

31.7hTotal Runtime

411.9MTotal Tokens

5,032 Tool calls

Run Results

Platform runs for the selected PoC setting. Success requires a verifier-confirmed primitive.

Platform	Success	Primitives	Runtime	Tokens	Tools	Date
Linux	20.0%5/25	`LEAK` 1/9`HREAD` 1/6`RIP` 2/6`HWRITE` 1/4	13.4h	202.0M	2,970	2026-04-26
Windows	5.0%1/20	`AAW` 1/20	21.7h	348.2M	3,845	2026-05-08
Linux	48.0%12/25	`LEAK` 4/9`HREAD` 4/6`RIP` 2/6`HWRITE` 2/4	9.9h	123.7M	1,929	2026-04-17
Windows	35.0%7/20	`AAW` 7/20	21.8h	288.2M	3,103	2026-05-08