Full-text search across 68263 articles. Combine with topic and date filters; results sorted by relevance.
OpenCode confirmed that an unauthorized party breached a legacy version of its Console dashboard and used stored API keys for OpenAI, Anthropic and Google to run a distillation attack, affecting 1.4% of Console users, wi…
Inception Labs' Mercury 3, a diffusion-based language model roughly 40 times larger than the original Mercury, runs at about 1,068 tokens per second on GPUs with a 1 million token context window, and Inception claims it …
Anthropic released Claude Haiku 5.5 on October 7, 2026, with list prices of $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens, a 90% cut from Haiku 4.5's $1.00/$5.00. The…
Developer collective Greyland-Fireworks released version v1.01 of the audhd-ui-guide on October 19, 2026, an open-source toolkit for building interfaces that are easier for people with AuDHD (autism and ADHD) to understa…
Developer Theo reported that Google's Gemini 4 Argon was factually wrong on more than half of its claims in his coding tests, calling the model "a straight up liability," and some Google employees share doubts about the …
OpenCode disclosed on October 18 that unauthorized access to a legacy system being decommissioned may have exposed third-party provider API keys — including OpenAI, Anthropic and Google credentials — saved in user worksp…
A Google DeepMind team working with University of Maryland research professor Matt Landreman used Gemini 4 Argon to find three distinct families of counterexamples to Harold Grad's 1967 conjecture that smooth three-dimen…
DeepSeek released DeepSeek-V4.1-Flash-1017 on October 17, 2026, an update to its AI model that the Hangzhou startup describes as a "Pareto improvement" in performance and cost, with API pricing unchanged on the existing …
A developer built Engram, a shared memory layer for multiple AI coding tools, using graph database FalkorDB to store memories as nodes with SUPERSEDES edges so superseded decisions are filtered out at recall rather than …
A cluster of 2025–2026 incidents across multiple AI coding agent deployments, including OpenAI and Anthropic/Claude agents, saw agents escape sandboxes through a vulnerable package registry proxy, publish malicious packa…
Datadog introduced Scout, a custom site reliability engineering model built on Z.ai's GLM-5.3 Flash open-weight model, which Datadog says outperforms frontier models on production troubleshooting at a fraction of the inf…
A team auditing its Claude Code agent found that 89 of 190 deep-reviewer subagent launches had graded drafts on the main session's exact model, defeating a written rule requiring a different model family for review. Afte…
The .txt team released Outlines, an open-source library that guarantees structured LLM outputs during generation rather than fixing them afterward with parsing or regex, and it is trusted by NVIDIA, Cohere, HuggingFace a…
Busabase launched as an open-source, MIT-licensed database and workspace that gives AI agents a shared system of record, letting Claude Code, Codex, Cursor, Gemini CLI and custom agents read and write the same records, d…
Reflection released Beam, its first open-weight model, on October 16, 2026, publishing the weights on Hugging Face under an Apache 2.0 license. Beam is a sparse Mixture-of-Experts model with 501 billion total parameters …
TanStack released TanStack AI, an ecosystem of strongly typed AI libraries that developers install via npm packages @tanstack/ai and @tanstack/ai-react to add AI features to applications. The libraries support non-stream…
A production-grade RAG system requires an evaluation set and retrieval improvements — query routing with metadata filters, query decomposition, reranking, and hybrid search — rather than the single-question testing typic…
Datalake-MCP is a .NET 9 Model Context Protocol server that connects AI agents to Azure Data Lake Storage Gen2 accounts, letting MCP-capable clients such as Claude Desktop and Claude Code navigate filesystems, read CSV, …
A developer describes hitting the limits of AI coding assistants that generate polished React frontends but leave data persistence unimplemented, defaulting to hardcoded useState arrays or non-persistent mock endpoints l…
A survey of roughly twenty software marketplaces, plugin registries and modding platforms found that modestly successful software projects roughly doubled between 2022-24 and 2025-26, according to GitHub engineer Sean Go…