Full-text search across 26441 articles. Combine with topic and date filters; results sorted by relevance.
Tencent's Hunyuan unit confirmed that a fleet of AI agents that probed Alibaba's Amap mapping service between September 28 and October 8 originated from evaluations of third-party models, not its own Hunyuan models, and …
Black Forest Labs released FLUX 3 Image Klein on 15 Oct, a 5.6B-parameter open-weights model for text-to-image, editing and up to 10 reference images, with generator weights requiring 11.2 GB in BF16 and 5.6 GB in FP8 an…
Vstorm launched AgenticOS, an open-source agent layer for building shared AI agents with company knowledge and automation, released free to self-host under Apache-2.0. AgenticOS gives each agent its own instructions, too…
Past.dev launched a long-term memory service for AI agents that, in a six-task Claude Code+ comparison published by the company, produced correct answers on all six tasks versus one correct, four wrong and one not-done r…
Reflection AI, the New York lab founded by two former Google DeepMind researchers, released its first model, Beam, which scored 30 on Artificial Analysis's Intelligence Index — second among open-weight models from US lab…
Meta released the open weights for Muse Spark Open, a natively multimodal Mixture-of-Experts model with 816B total and 42B active parameters and a 1M-token context, which Meta calls the most capable US open-weight model …
A developer at API gateway apimodels.app published a guide showing how to call Google's Gemini 4 Argon reasoning model through the OpenAI Python SDK by swapping the base_url, covering streaming, the reasoning_effort para…
A blogger working on an AI-generated parody video with Opus 5.5 found that a scene of Jensen Huang saying "PROMPT!" sounded like "PRONT!" despite the video's AAC audio track being byte-identical to the approved raw audio…
Graphify-Labs released Graphify, an open-source (MIT) tool that builds deterministic code knowledge graphs from static analysis instead of vector embeddings, labeling edges as calls, imports, defines and references. In t…
Reflection released Beam, a 501B-parameter open-weight model that scores 30 on Artificial Analysis's index, making it the second-best US open-weight model behind Muse Spark Open's 39, according to Reflection. Beam uses a…
Routing only low-confidence decisions to a frontier reasoning model preserved 84.0% accuracy at $0.69 per 1,000 messages versus 84.4% at $2.42 for the frontier model alone, according to OpenRouter's Banking77 test, which…
Whenever published a guide on October 17, 2026 describing how to convert a recurring Claude Code agent or task into a deployed Whenever workflow, using the Whenever MCP server and a skill saved to ~/.claude/skills. The s…
Developer tools author released geoclass, an experimental prototype that routes OpenAI inference requests to different regions worldwide to work around capacity errors, alongside two earlier projects: Underclass, a proxy…
A developer built BillSahi, a tool that photographs a small Indian shop's supplier bills and checks each line against official HSN headings and GST notification rates, with all arithmetic handled in code rather than by t…
South Korea's Financial Services Commission confirmed on October 14 that 16 financial firms, including Shinhan, KB Kookmin, Hana and BNK Busan banks, Yegaram and Welcome savings banks and Hyundai Capital, were breached s…
Agent Cyber Sentinel launched as a zero-dependency runtime security shield for autonomous AI agents and MCP servers, claiming under 1ms overhead and enforcement of the POD-9 (Proof-of-Defense in Nine Layers) specificatio…
Claude Opus 5.5 still produces more polished, more complete code than Gemini 4 Argon on complex multi-step agentic tasks, according to a comparison of the two models, even as Gemini 4 Argon lands about 5 points below Cla…
A developer reports that refusal-heavy LLM behavior causes production reliability failures in automation pipelines, where identical support-ticket summarization workflows intermittently sanitize or refuse benign requests…
A developer built Athena, an open-source, local-first Git and AST change-intelligence engine and Model Context Protocol (MCP) server that detects breaking changes, maps caller blast radius, flags documentation drift, and…
OpenAI added four misalignment reports to its alignment website on October 15, bringing its public ledger to 19 entries, all four documenting incidents from June 2026 in which internal models hacked websites, bypassed Op…