AI Research news and analysis on Web Pulse: 21288 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
A study posted to arXiv, MemToC, found that instruction-tuned language models retained a correct answer only 6.5% to 17.1% of the time when an external tool returned incorrect information, with results pooled over three …
Amazon Web Services released 38 open-source agent skills spanning 11 healthcare and life sciences domains under the MIT-0 license, encoding domain decision procedures in SKILL.md markdown files that agents consume at inf…
NVIDIA's Resiliency Extension (NVRx) can be integrated into PyTorch Fully Sharded Data Parallel training on Amazon Elastic Kubernetes Service to cut idle GPU time, with synchronous checkpointing consuming up to 40% of to…
An independent developer released a reverse-engineered, open starter model called Jev-like that scores a changing list of N text options in a single pass, matching the input and output shape of TypeSafe's unpublished com…
A team of researchers created their own contemporary model papyrus scrolls, charring them like the originals, to validate a screening method for determining which of the roughly 600 Herculaneum scrolls were written in le…
Google researchers built the first portable vectorized Quicksort, using Highway's portable SIMD functions to run on six instruction sets across three architectures, and reported a 9-19x speedup over the standard library.…
Artificial Analysis's Intelligence Index version 4.1.1 weights agentic workloads at 34% and general reasoning at 18%, a shift an analysis of 586 model evaluations argues distorts what the score measures. The analysis fou…
GoBench, a new benchmark from researcher Roland Gao, evaluates frontier large language models on 9×9 Go against a calibrated ladder of KataGo opponents used as Elo anchors, with an arXiv release scheduled for September 1…
Rijul, a developer building the AI code review tool LiveReview, outlined the difference between autoregressive and diffusion-based text generation. Autoregressive models, used by most deployed language models today, gene…
A developer analyzed 28,721 real repositories and found that the median AI instruction file contains about 50 items, only a dozen of which are actual directives, with the rest being scaffolding that competes for the mode…
A developer argues that AI safety should be modeled on human institutions rather than a single perfectly aligned reward function, proposing that a society of agents with roles, monitored enforcement, and formal computati…
NVIDIA's Vera Rubin NVL72 delivered up to 3.7x higher throughput than the GB300 NVL72 on the Qwen3-VL model in MLPerf Inference v6.1 preview submissions, using vLLM with the NVIDIA Dynamo framework, while DeepSeek-R1 thr…
GPT-6 Astra saturated ArXivMath at 94% and BrokenArXiv at 96%, prompting benchmark operators to revise both pipelines so that GPT-6 Astra now scores 88% on ArXivMath August and 81% on BrokenArXiv August. All models are n…
8Braid reproduced OpenAI's September 8 AI-generated proof of finite-time breakdown for the Navier-Stokes equations and its Lean formalization, confirming that both the Nanoda kernel and Lean's default kernel accepted the…
OpenAI's GPT-6 Astra topped the Epoch Capabilities Index with a composite score of 166, outranking all 249 models tracked by Epoch AI and surpassing GPT-5.5 Pro at 162 and Anthropic's Claude Fable 5.1 at 164. Astra score…
Epoch AI's AI Data Centers database now tracks 86 large AI facilities worldwide, covering an estimated 46% of all deployed AI compute as of June 2026, up from roughly 22% in early 2024. The tracked sites hold a combined …
OpenAI capabilities researcher Dan Selsam, who has worked at the company since 2022, issued a public statement on AI risk shared via a tweet by AI 2027 author Daniel Kokotajlo, pointing to increasing model situational aw…
A new system called RelateAnything performs real-time open-vocabulary relation prediction from any inputs, according to the research announcement. The work targets a gap in scene-graph modeling, where models remain train…
A proposed shared selective persistent memory architecture for agentic LLM coding systems filters four categories of reusable context — task specifications, data schemas, tool-use patterns, and configuration choices — in…
AI compilers differ fundamentally from classical compilers because they operate over a small set of recurring model families — transformers, convolutional networks, diffusion models, and mixture-of-experts — and because …