Full-text search across 67594 articles. Combine with topic and date filters; results sorted by relevance.
A computer science student built a four-agent LLM pipeline (Planner, Coder, Reviewer, Explainer) on Groq's LLaMA 3.3 70B and benchmarked it against two single-agent baselines on 50 ground-truth data-analysis questions ac…
Artificial Analysis rebuilt its Intelligence Index so that 60% of it now runs on private or held-out test sets, and Claude Opus 5.5 leads the new ranking at 58, one point ahead of GPT-6 Astra's 57. The rebuilt index adds…
JFrog Security Research reported Monday that attackers compromised 42 third-party GitHub repositories by deploying payload blobs as large as 13.7MB to evade the GitHub search and API paths normally used to detect such co…
Google released Android Studio Rabbit 2 as a stable production release, adding Bring Your Own Agent support for more than 40 ACP-compliant coding agents, including Claude Agent, OpenAI's Codex, Junie, and Google Antigrav…
OpenCode 2.2.0, published 27 Oct 2026, opened its desktop app to createable, installable and shareable plugins — marked experimental in the release notes — that can add keyed session instructions without editing the syst…
Amazon engineering leadership posted a revised software engineer role guideline on aboutamazon.com on October 27, 2026, telling engineers to prioritize shipped outcomes over AI tool usage and calling token consumption, A…
A developer has proposed a vendor-neutral typed error envelope for agent tool results, designed to replace ambiguous human-readable error strings with machine-readable fields that tell the runtime whether a state change …
Carbon-X, a 14.8B-parameter open-weight DNA language model built on Qwen3-14B, averaged 66.7% accuracy across five DNA benchmarks with reasoning disabled, just below GPT-6.1 Sol's 67.5% and ahead of Opus 5.5's 60.9%, at …
A developer built Aitoword, a free browser-based converter that turns AI-generated Markdown and LaTeX into editable .docx files entirely client-side. The tool isolates code blocks before normalizing math delimiters, pars…
Microsoft Research published Flint-interactive on October 22, 2026, adding an interactive layer to Microsoft's MIT-licensed Flint chart compiler, which turns short chart specs into native specs for Vega-Lite, ECharts, Ch…
SparkLLM released Spark-X2.6-4B, a 4-billion-parameter open-weight model under an Apache 2.0 license that the team's model card claims matches or beats some 9B-parameter models on coding and agentic benchmarks while runn…
Goodfire published a research essay arguing that neural networks are far more interpretable than the field has assumed, with researcher Nick Cammarata reporting that 99 to 100 percent of neurons in the final layer of an …
A developer's analysis of the Stack Overflow 2026 Developer Survey found SQL usage at 58.4%, statistically tied with Python at 58.0% and HTML/CSS at 58.2% given a roughly ±0.8-point margin on about 14,099 respondents. Th…
A developer built four open-source tools spanning a seismic data pipeline: torch-friction, a differentiable PyTorch rate-and-state friction layer that recovered a, b and Dc parameters with under 0.05% error after inversi…
Developer vicmaster ported a panel from coide, their self-built Electron GUI for Claude Code, into skill-atlas, an open-source Claude Desktop mod built in a day and installed via a symlink. The mod is a read-only pane wi…
Vlt released version 1.5.0, adding a `vlt skills` command that discovers agent skill folders shipped inside installed JavaScript dependencies and links them into the directories coding agents such as Codex, Cursor, GitHu…
Amytis, a data-analysis tool from co-founder and CTO Freddie Starkey, generates Python and R scripts from plain-language instructions through its Python Factory and R Factory cards, then runs them on demand. The tool fee…
A developer released Toktail, an Apache 2.0-licensed open-source proxy that sits between coding agents and AI providers to count input and output tokens per model, host and day, logging each request as a CSV row and mirr…
Spirit Room launched a service that tests and rewrites MCP server, CLI and API tool descriptions so AI agents complete tasks in fewer steps and with fewer errors, claiming a pilot takes about two weeks with zero code cha…
The open-source project tokenbud released a desktop app that lets a host share access to a local or API-backed AI model through a 9-digit ID and password, without handing over the underlying API key. Hosts set a cap on c…