A preprint describes a vision-language-action model that learns compact physical tokens about how scenes will change, sidestepping the delay of generating video.
A video world model from Tencent Hunyuan follows successive typed instructions and camera moves while keeping track of what has already happened on screen.
arXiv preprint · 27 October 2026
A KAIST AI preprint steers video generation with geometry maps rebuilt from reference photos, so a subject stays the same as the camera circles it.
arXiv preprint · 26 October 2026
A preprint offers a more realistic test for audio-visual voice extraction, plus a streaming model the authors say is far smaller and faster than an existing baseline.
arXiv preprint · 23 October 2026
A decoding method from KAIST AI checks whether the picture would change the answer before processing it, cutting reported decoding computation to about a third.
arXiv preprint · 27 October 2026
Sequencing more than five million immune cells from 1,925 people shows how inherited variation tunes gene expression, often in just one or a few cell types.
Nature · 28 October 2026
A preregistered analysis of 40 brain-imaging datasets finds that common measures of how well people reframe their feelings disagree, and none tracks prefrontal activity.
Nature · 28 October 2026
A study ties genetic variation to single-cell chromatin accessibility and finds that the genetic signal helps an AI model map how genes are switched on.
Nature · 28 October 2026
Stimulation and recordings in 87 neurosurgical participants map four functional regions of the insula and an asymmetric pathway through which cognition may shape pain.
Nature · 21 October 2026
A large-scale evaluation suggests that the way memory systems reorganise an agent's history can hide evidence, a cost the authors call memory tax.
arXiv preprint · 25 October 2026
A test built on hundreds of synthetic sessions finds the best agent memory method answering only about a third of habit questions correctly.
arXiv preprint · 20 October 2026
In a benchmark where diagnosticians must decide what information to gather, a language-model system far outscored generalist doctors, under controlled conditions.
Science · 22 October 2026Microscopy reveals two localised electronic states hiding among a metal's mobile electrons, an electronic counterpart to an effect already achieved with light.
Science · 22 October 2026
Polarised X-ray scattering uncovers a magnetic ripple with a definite twist, whose spin polarisation varies with momentum in line with the material's symmetry.
Science · 29 October 2026
A preprint describes a training method that replays a robot's decisions in a learned simulator, tries alternatives, and credits the choices that made the difference.
arXiv preprint · 26 October 2026
A preprint names a training pitfall called interface lock-in and reports that replaying mixed spatial data in time order improves video models on three benchmarks at once.
arXiv preprint · 25 October 2026
A preprint describes one shared encoder and one output schema for ten parsing tasks, with the authors reporting leading scores on document and multimodal benchmarks.
arXiv preprint · 29 October 2026
TAO, described in a preprint, picks which agent sessions get GPU time by weighing when their tools will return, what is cached where and how many tokens they carry.
arXiv preprint · 27 October 2026
A preprint from Xiaomi Research describes a training loop that watches its own learning signal and swaps in fresh, checkable questions, with no human-labelled data.
arXiv preprint · 28 October 2026
ThinkWorld, a preprint framework from NVIDIA-affiliated researchers, rewards a robot model when the imagined outcome of a planned motion matches what real footage shows.
arXiv preprint · 27 October 2026
A preprint proves that one family of AI models learns the same representation across data types when trained on the right pairs, and calls next-token prediction an inefficient route.
arXiv preprint · 29 October 2026