Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-08-07 · high
ByteDance Seed's GST-Bench, from 6,790 minutes of video, shows models track local space fine but lose the global picture.
Why now: The GST-Bench paper posted to arXiv and made the August 7, 2026 briefing.
GST-Bench catches vision-language models losing track of space over long video, per ByteDance Seed's paper on arXiv.
The gap is stark: top models average 42.68 on the benchmark's spatial consistency tests, against 79.08 for humans, a roughly 36-point spread. That matters most for agents that operate over long video, since a wrong mental map compounds with every step that follows it.
The benchmark draws on 6,790 minutes of synthetic video with human-verified questions, built to isolate one specific failure. Models handle local spatial relations, judging what sits near what in a single view, without much trouble.
The failure shows up when models have to stitch separate observations into one consistent scene as an episode stretches on.
ByteDance Seed also released GST-Train, a training set built to target that consolidation problem directly.
Whether GST-Train closes the 36-point gap is the thing to watch next. If it doesn't, the failure sits deeper than a data problem, in how these models represent space at all.
Each link below shares sources, entities, or timing with this story.
Shared entity: Bench / Same source domain / Shared topic / Earlier coverage / Tension
Both cover Bench; reported by the same outlet (arxiv.org); overlapping topics (agent, failure).
Both cover Bench; reported by the same outlet (arxiv.org); overlapping topics (against, agent).
Shared entities / Same source domain / Earlier coverage
Both cover Bench, Train; reported by the same outlet (arxiv.org); earlier Bench coverage from 2026-07-16.
Shared entity: Bench / Same source domain / Shared topic / Earlier coverage
Both cover Bench; reported by the same outlet (arxiv.org); overlapping topics (agent, attack).
Both cover Bench; reported by the same outlet (arxiv.org); overlapping topics (against, attack).
Both cover Bench; reported by the same outlet (arxiv.org); overlapping topics (agent, attack).
Shared entity: Bench / Same source domain / Earlier coverage / Tension
Both cover Bench; reported by the same outlet (arxiv.org); earlier Bench coverage from 2026-08-06.
Both cover Bench; reported by the same outlet (arxiv.org); earlier Bench coverage from 2026-08-03.