WorldMemBench (zju-real/worldmembench) is a benchmark that evaluates the ability of video world models (video world model) to remember scenes. The commit 1b79840 of October 15, 2026 is a documentation change that touched only one file in this repository, README.md. The commit message is "docs: update paper links, release news, and citation", and 32 lines changed in the README (27 lines added, 5 lines deleted).
The center of this change is the badge line at the top of the README and the newly added News and Citation sections. The arXiv badge now points to arxiv.org/abs/2610.19063, and a Hugging Face Daily Papers badge was added. The existing Hugging Face Leaderboard badge, the GitHub Code badge, and the MIT license badge disappeared from the badge line, and instead a Leaderboard Results badge pointing to the #leaderboard anchor of the project site came in. The newly created News section records the arXiv paper release on October 13, 2026, the release of the evaluation code on October 9, and the uploading of inference results to Hugging Face on the same day. A BibTeX block containing the paper title and author list, the year 2026, eprint 2610.19063, and primaryClass cs.CV was added to the Citation section.
However, the file touched by this commit is just the single README, and there are no code changes. The license indication has only disappeared from the badges; the README's License section states that the MIT license applies to the code, that separately distributed datasets and examples carry their own terms, and that third-party submodules and model weights retain their original licenses. The dates and result list in the News section are records the repository itself wrote, not independently verified figures, and the Leaderboard badge merely links to a site anchor, so the detailed comparison figures must be found in the body of the README. At this stage, it is appropriate to regard this as the point at which the paper and the evaluation code have been made public, and whether it reproduces as-is in other environments is difficult to judge from this change alone.