Full-text search across 70329 articles. Combine with topic and date filters; results sorted by relevance.
A developer built GHOST (Generative Human-memory Offline Storytelling & Tracking), an open-source, fully local memory archive that lets families store letters, photos, recipes and transcribed oral histories and query the…
Vercel Container Registry now accepts compressed image layers up to 2 GB, up from the previous 500 MB limit, while the total image size limit stays at 15 GB. Vercel said the higher per-layer cap lets developers push imag…
A developer's cryptic crossword solver, ximenes, built over fourteen pre-registered briefs with three coding agents, filled 17 of 40 post-cutoff Guardian, Independent and FT grids with every answer correct, but a single …
A developer tested an existing BrowserAct Semrush Bot as an MCP tool inside OpenAI's Codex, requesting a US Keyword Overview for "OpenAI Codex" and receiving one keyword record (14,800 estimated monthly searches, difficu…
Australia's House of Representatives passed the Online Safety Amendment (Digital Duty of Care) Bill 2026 on October 28, sending it to the Senate with penalties of up to approximately $109.2 million for corporations that …
OpenAI has deprecated the POST /audio/translations endpoint, according to the company's published API specification. The endpoint still functions today, but the deprecation notice is an early warning that it will eventua…
Anthropic reported on October 28, 2026 that it is investigating elevated errors when searching and reading email through the Gmail connector, a fault the company said does not affect other connectors. The incident, first…
A research team posted arXiv:2610.36249v1, introducing HiRe-DRR, a Hierarchical Multimodal Interaction Reasoning framework for dyadic relationship recognition that outperforms recent image-, video-, and multimodal method…
Researchers introduced SOMA, a benchmark of 1,632 human-verified questions for fine-grained egocentric body-motion understanding, spanning body configuration, motion dynamics, and temporal reasoning. In evaluations of 37…
A new arXiv paper, BoxWorld (arXiv:2610.36371v1), separates interaction reasoning from generative rendering in real-time interactive video worlds by having a coding agent build a "Box World Engine" from user-specified ru…
Researchers released DocXEval, described as the first comprehensive explainability framework for multimodal document understanding, evaluating 13 XAI methods across 3 tasks, 8 datasets, 4 models and 17 evaluation metrics…
Typed decision models served by a single commercial inference gateway classified knowledge conflicts in RAG at 50.0–67.0% accuracy, and 10 of the 11 matched or beat Gemini 2.5 Flash's 61.3% at 4–39x lower billed cost and…
A team of researchers posted arXiv paper 2610.36325v1, an ontology-guided multi-agent framework for differential diagnosis that grounds LLM diagnostic inference in a structured, expert-curated medical knowledge base to r…
A new arXiv paper (2610.36437v1) proposes CONCUR, an advantage estimator for multi-task, multi-reward reinforcement learning that replaces the standardized scores of each gated auxiliary reward with the closest scores sa…
A new arXiv paper (2610.36006v1) evaluates MedJev, an open "System One" model that runs locally on a laptop and answers structured medical questions in a single forward pass, reporting accuracies of 72.5% on MedQA, 75.4%…
Researchers at the Tang Lab at the University of British Columbia released Neural-DISCO, a disentangled representation learning framework that partitions latent space into label-specific components plus a residual latent…
A new arXiv paper (2610.36511v1) introduces ClaimNov, a claim-level novelty module for AI scientific discovery pipelines that catches 97% of disguised papers at a 1.5% false-alarm rate and cuts an LLM judge's false alarm…
A new arXiv paper (2610.36353v1) proposes a label-preserving perturbation framework for detecting ambiguous inputs to large language models before the models attempt to answer them. The authors instantiate the framework …
Lossless LLM compression schemes such as DFloat11 can cut tensor data volume by an average of 1.4x while retaining all information and preserving model accuracy, offering an alternative to quantization, which degrades ac…
Multipeer released version 0.6.0, adding git worktree support so several agents can work in one project without conflicting, with each agent card showing its branch. Agents now name themselves after the first prompt on C…