Full-text search across 71119 articles. Combine with topic and date filters; results sorted by relevance.
Ultramarine 1 Flash, an agentic model built by applying large-scale reinforcement learning to the open base model GLM-5.3 Flash, improves on that base across all six knowledge-work benchmarks the company evaluated, by 4.…
Agentic policies that write and iterate on their own Python skills reached a 34% to 65% overall task success rate across ten bimanual tasks, beating end-to-end policies trained on 30 times more demonstrations on 8 of 10 …
Thinking Machines released Inkling 2 Small, a 594B-total, 17B-active multimodal mixture-of-experts model pretrained from scratch in under 3 days on fewer than 12T tokens in native NVFP4, and it is available now on Tinker…
A working implementation of Cisco's published Foundry security harness specification was pointed at Langflow 1.7.3's entire v1 API — 289 function-level units across 25 modules — as part of a Security Loop Series experime…
A developer built "Do I Need to Do This?", a web app in which an open Gemma 3 model running locally via Ollama gathers facts about a task and a roughly 200-line plain JavaScript rules file makes the final verdict, so the…
Over roughly 48 hours on 10/21/2026-10/22/2026, a cluster of critical vulnerability disclosures and breaches that AINews labeled "Black October" hit defenders at once, including an unauthenticated RCE in PHP dubbed PHP2S…
Anthropic replaced Git on its giant monorepo with a new server-side source control system called Scion, built by a small team with heavy use of Claude over about five months, after the monorepo's merges grew from 2,500 a…
Earendil released @earendil-works/pi-durable at v1.1.0 on October 7, 2026, an experimental runtime that replaces the in-memory while-loop pattern of AI agents with an embedded SQLite storage engine and an operating syste…
Anthropic's Lance Martin published "Automating eval design and hillclimbing with Claude" on September 28, adding two commands to the claude-api skill: /claude-api build-eval, which interviews the user and builds an evalu…
ProjecturEd, a Julia-based application for viewing and editing structured data through multiple editable projections, demonstrated a local AI assistant building an M/M/1/K queue study in a video where the assistant typed…
Spectro Cloud reported that 85 of its engineers processed 41 billion tokens in a one-month pilot of its PaletteAI Inference Launchpad, with 40 billion tokens handled locally on a single server with eight NVIDIA B200 GPUs…
Google limited Gemini 2.5 model access in its API on September 18, 2026 to users who have actively used the models in the past, stating the 2.5 models are not deprecated and will continue to be served until further notic…
Developer miruky published a walkthrough using a single three-line Python duration parser bug to distinguish prompt, context, harness, loop, and graph engineering, showing how each layer changes what is asked of a model …
A review of documentation for seven AI admin consoles — from OpenAI, GitHub, Anthropic, Microsoft, Google and Cursor — catalogued 85 named metrics and found that all seven count seats and active users, but no two define …
A 2026 guide details how LLM model routing middleware can cut API costs by 40–85% without measurable quality loss by dispatching each request to the most suitable model. It outlines three architectural patterns — fallbac…
OpenAI's GPT-6-Astra model can lay out a printed circuit board in KiCad, build a 3D city scene in Unity, create an animated automobile transmission in FreeCAD and Blender, and draft a tax return from a W-2, according to …
A community-maintained handbook compiles ChatGPT slash commands, UI shortcuts, Canvas triggers, and custom command engines into a single reference, noting that OpenAI publishes no official master list. The guide categori…
A developer compared three approaches to connecting tools in the Claude API: traditional JSON-schema tool calling, the SDK's Tool Runner helper, and programmatic tool calling. The comparison found that traditional tool c…
Kong and Straiker announced an integration that combines Kong AI Gateway's traffic enforcement with Straiker's agent-level security analysis, including Straiker's Agentic Kill Switch, which can revoke tools, suspend sess…
A developer built an architecture connecting Google Cloud Run Sandboxes (gVisor) directly to Google Apps Script, enabling deterministic Python and Bash execution inside Google Workspace in 200–450 ms with zero-trust micr…