Full-text search across 12727 articles. Combine with topic and date filters; results sorted by relevance.
Researchers introduced probabilistic language tries (PLTs), a unified framework that represents prefix structures from generative models, enabling optimal lossless compression, policy representation for decision-making, …
Google Data Cloud's frontier AI team introduces Discovery Bench, a meta-benchmark that uses information theory to modulate evaluation case difficulty for AI agents, moving beyond pass/fail metrics to map agent capabiliti…
Scientists developed a label-efficient AI model using Siamese neural networks that predicts phase transitions in physics with minimal data, achieving accuracy on par with established methods without requiring critical re…
Researchers introduced ALER-TI, a novel time series imputation method that uses historical data patterns to reconstruct missing values more accurately than existing approaches. By aligning latent embeddings between incom…
OpenAI launched the GPT-5.6 model family (Sol, Terra, Luna) and merged Codex into the ChatGPT Work desktop app, introducing programmatic tool calling and parallel agent coordination. The shift from passive code generatio…
AI companies captured 65% of US venture dollars in 2025, but the funding is concentrated in megadeals while small seed rounds shrank 20%. Only about one in six seed-stage AI startups reaches a Series A within two years, …
Efficient Long-hOrizon (ELO) learning, a new learned optimizer, surpasses traditional models like AdamW and Muon across diverse tasks including language modeling and image classification, requiring less than 7 H100 GPU-h…
A developer built a Double-Tower Gated Transformer Network (GTN) to generate profitable trading signals from high-frequency EUR/USD forex data. The system uses a custom cost matrix and weighted cross-entropy loss to comb…
Keysight Technologies launched the APS-ONE-400, a modular network cybersecurity test platform delivering 400 Gb/s Layer 4-7 throughput and 380 Gb/s encrypted TLS throughput, designed to handle elephant flows from AI work…
Chrome 149's WebMCP origin trial is now open for registration, and Microsoft Edge 147 has shipped native WebMCP support by default. WebMCP is a client-side protocol that allows websites to declare what AI agents can do w…
The agentic web is splitting into two bets: identity via llms.txt and capability via WebMCP. Llms.txt, a file that tells AI models what a website covers, is being enabled by default on millions of WordPress sites despite…
Blackthorn Vision, a Microsoft-partnered .NET and AI development company, replaced a custom prompt orchestration layer with Semantic Kernel in a production healthcare SaaS product. After six months, the team found that S…
Crawl4AI, an open-source web crawler and scraper that converts web pages into LLM-ready Markdown, has released version 0.9.1 with 12 bug fixes. The tool, which has over 50,000 GitHub stars, is designed for RAG, agents, a…
Developers building real-time voice AI applications must abandon synchronous tool loops to achieve sub-second latency, according to engineering patterns from Ello and Khan Academy. Streaming action interpreters that pars…
A new review of remote sensing foundation models (RSFMs) finds that no single geospatial foundation model is universally best and that inconsistent evaluation hinders fair comparison and reliable deployment. The chapter …
Researchers developed a graph neural network model for real-time hand gesture recognition using surface electromyography (sEMG) signals. The model achieved 99% classification accuracy on data from 8 subjects using a Myob…
A new survey from arXiv explores post-training techniques for end-to-end autonomous driving models, addressing limitations of open-loop imitation learning in safety-critical environments. The paper categorizes existing m…
OpenAI's GPT-5.6 family includes three tiers—Sol, Terra, and Luna—with Sol reportedly used to train Luna, the cheapest model, according to Array Ventures founder Shruti Gandhi. The pricing strategy positions Luna as a hi…
Context.dev, a Y Combinator-backed startup, launched an API that uses LLMs to perform schema-driven web scraping, allowing developers to extract structured data from any website by defining a JSON Schema instead of writi…
At least 22 professors and researchers from Stanford, Berkeley, and Harvard have left or taken leave in 2026 to join OpenAI, Anthropic, Google DeepMind, and Meta, accelerating a brain drain from academia into corporate A…