Articles & Guides

Researchers from DeepMind and collaborators have redesigned the optimization problem behind combination loss analysis and refined the search with machine learning and AlphaEvolve. The work improves the known upper bound on the matrix multiplication exponent from 2.371339 to 2.371177.

Ventor-QTest audits hosted LLM inference without requiring token probabilities from the provider. It separates average output drift from rare, severe failures in long sequences.

The DumpsterCluster project shows that retired GPUs can be assembled into a practical large-model inference platform. Its low hardware cost comes with a major condition: electricity prices and grid carbon intensity determine whether the system is truly economical or sustainable.

An audit of 37 language models finds that plausible synthetic survey answers can reproduce broad human relationships while failing to preserve the deeper structure of real respondents’ data.

Poisoned retrieval contexts can make a language model appear more confident rather than more uncertain. A new study proposes D-SCAN, which detects attacks by tracking document-level attention collapse during generation.

UI-Mate combines an environment-grounded training pipeline with in-context demonstration learning for long-horizon computer tasks. Its 27B model reaches 77.0% on OSWorld-Verified, according to the supplied paper summary.

A paper from Shanghai AI Laboratory proposes a three-level framework for understanding risks created by the expanding cognitive scope of LLM-based agents. It asks whether humans can preserve agency, autonomy, and meaningful control as agents move beyond text generation.

WorldRover introduces an Unreal Engine-based pipeline that renders long exploration routes together with geometry, motion, and correspondence annotations. The resulting data is designed for models that must maintain and revisit coherent representations of interactive worlds.

A study of real receipt fields finds that ordinary confidence-thresholding can miss its risk target because document fields are clustered, scores leak into threshold fitting, and discrete scores create unstable thresholds. It proposes a validity ladder and identifies when conditioning improves coverage rather than merely fragmenting the calibration set.

SPARGen reframes several spatial intelligence tasks as instruction-conditioned generation within a native multimodal model. Its main contribution is a unified interface where geometric, correspondence, and reasoning supervision can shape shared representations.

OpenAI is adding Computer History to the ChatGPT desktop app on macOS, letting ChatGPT and Codex reference a timeline of user actions. The feature avoids screenshots, but it still raises familiar questions about how much of a personal computer an AI assistant should be allowed to remember.

A security-focused investigation describes how unused AI credits are being bought from startups and resold through brokers, marketplaces, and discount routing services. The emerging trade turns inference capacity into a quasi-currency with cost, compliance, and abuse implications.

UniSwap proposes a unified framework for replacing both the visual identity and vocal timbre in talking videos. Instead of stitching together separate face and voice models, it performs joint audio-visual transfer inside a single diffusion transformer.

LiveAnimate targets a familiar tradeoff in pose-driven human animation: diffusion models can look good, but they are too slow and too memory-hungry for interactive use. The paper reframes a 14B video DiT into a streaming system built for stability over long sequences.

A new paper turns “staying coherent over many turns” into an automatic benchmark. NCP-Bench shows that even strong models can drift, contradict themselves, and lose track of key story commitments.

StateFlow proposes a state-centric framework for generative previsualization across film, games, architecture, and urban design. Instead of producing a video in one shot, it maintains an editable 3D world that can be constructed, evolved, and accessed over time.

Ex-Omni-2D tries to close a gap in today’s omni-modal assistants: they can talk, but they still feel visually absent. The paper combines text, personalized speech, and reference-conditioned video into one dialogue pipeline.

VibeLifeBench reframes agent evaluation around long-running everyday tasks where the world changes silently and constraints are often implicit.

UniMoMo is a post-training compression method for large recommendation MoE models. It merges experts by functional behavior on calibration data and protects high-traffic routing paths to preserve quality while improving serving speed.

SWE-Bench ProMax shifts coding-agent evaluation from narrow bug fixing to large-scale, behavior-preserving refactoring across multiple files and languages.

Ouroboros reframes coding agents as systems that can improve not only outputs, but also their tools, prompts, context assembly, and core implementation. The paper reports leading results on Terminal-Bench 2.1, OSWorld-Verified, and CL-Bench while emphasizing governed evolution.