OpenAI has expanded Codex beyond code generation, enabling it to see and manipulate Windows graphical interfaces and be remotely orchestrated from the ChatGPT mobile app, signaling a shift toward AI agents that operate within real computer workstreams.
Claude Opus 4.8, released just 43 days after 4.7, improves honesty, cuts code‑defect miss rates to a quarter, reduces over‑confident answers, outperforms Mythos on several benchmarks, and introduces Dynamic Workflows that let hundreds of sub‑agents run in parallel for complex tasks.
The article presents RTPurbo, a lightweight two‑stage training method that converts full‑attention LLMs into highly sparse models with over 97% sparsity, achieving up to 9.36× prefill and 2.01× decode speedups while preserving near‑lossless accuracy across long‑context benchmarks up to 512K tokens.
WBench, a new benchmark created by Meituan LongCat and Fudan University, evaluates 20 state‑of‑the‑art video and world‑model systems across 289 test cases and 1,058 interaction rounds, measuring video quality, setting adherence, interaction fidelity, consistency and physical compliance, and reveals that no model yet excels in all five dimensions.
Anthropic unveiled Claude Opus 4.8, a faster, more honest LLM that improves benchmark scores across six of seven tests, introduces dynamic workflows for Claude Code, previews the higher‑tier Mythos model, and announced a $65 billion Series H round that lifts its valuation above OpenAI.
The SenseNova-U1-8B-MoT-Infographic model dramatically improves AI‑generated infographics by enhancing dense‑text rendering, layout stability, and chart accuracy through targeted data, extended mid‑training, and reinforcement‑learning fine‑tuning, achieving top scores on BizGenEval and IGenBench and surpassing many commercial rivals.
This article explains how QA engineers can use the AI‑powered Requirements Analysis Skill in TRAE IDE to automatically convert lengthy PRDs into prioritized module lists, GWT acceptance criteria, risk matrices and test checklists, and shows three ways to import the skill and three practical scenarios.
Anthropic’s Claude Opus 4.8 introduces Dynamic Workflows, letting Claude generate JavaScript orchestration scripts that schedule hundreds of sub‑agents, turning chat‑based plans into auditable code suitable for large‑scale audits, migrations, adversarial reviews, and long‑tail clean‑ups while exposing clear limits on concurrency, permissions, and token cost.
Anthropic's Claude Opus 4.8 adds modest performance gains, longer context, fast mode, effort control, dynamic workflows, and higher honesty, turning the model from a chat assistant into a dispatchable engineering squad—a shift that brings real‑world productivity benefits but also new risks for developers, product managers, and designers.
The article reviews the open‑source “The Agency” repository, which bundles over 170 AI‑agent subagents across 17 departments—from engineering and design to marketing and sales—providing role‑based prompts, SOPs, and deliverables for Claude Code and other tools, and shares installation steps, usage examples, and practical tips.
Integrating diverse AI agent harnesses into reinforcement‑learning pipelines is notoriously labor‑intensive, but NVIDIA’s new Polar system inserts an API‑proxy layer that treats any harness as a black box, enabling seamless rollout recording and trajectory reconstruction, as demonstrated by dramatic performance gains on a 4B model across multiple harnesses.
This work analyzes weight‑direction changes in diffusion‑model distillation, proposes a low‑rank rotation adapter (LoRaD) to model those changes, and integrates it into Variational Score Distillation as WaDi, achieving state‑of‑the‑art FID on COCO with only ~10% trainable parameters while generalizing to multiple downstream tasks.
ZhiYuan’s Genie Envisioner‑Sim 2.0 (GE 2.0) captured the overall WorldArena world‑model title without any task‑specific tuning, demonstrating superior long‑sequence stability, multi‑view generation, real‑time inference and a closed‑loop reward feedback loop that outperforms industry baselines across 16 metrics and three real‑world tasks.
After a team fine‑tuned a customer‑service model with internal complaint data, the article shows why AI‑driven data quickly becomes unaccountable, then presents a three‑step protocol—weight table, automatic log generation, and commercial‑benefit mapping—to transparently trace contributions and prevent ghost‑labor disputes.
The article explains that the hidden infrastructure layer called Agent Harness—its OS‑like architecture, three‑layer abstraction, context‑rot problem, compounding error, and verification loops—determines whether impressive agent demos can survive in production, with concrete benchmarks showing harness improvements far outweigh model upgrades.
The article explains how repetitive prompting of AI coding assistants can be eliminated by using the open‑source Skill mechanism, which packages project conventions and custom actions into reusable files, describes the two Skill types, placement options, and lists ready‑made Skill collections and tools for developers.
This article explains what CLAUDE.md is, why overly long files hurt Claude Code, how to write concise, verifiable rules, organize them hierarchically, use /init and /memory commands, and provides a practical template, backed by community data and Anthropic documentation.
Nous Research has released Hermes Agent v0.15.0 “Velocity”, an open‑source AI‑agent framework that consolidates 747 pull requests into a 4 500‑fold faster core, adds session‑sticky routing, new hermes send and audit commands, enhanced security, multi‑agent Kanban, and numerous integration and UI improvements.
The article presents Token Superposition Training (TST), which temporarily averages embeddings of non‑overlapping token bags and predicts groups of tokens in a first phase before reverting to standard token‑wise prediction, achieving up to 2.5× pre‑training speedup on 10B‑1B MoE models without altering model architecture or inference.
ControlFoley, released by Xiaomi's large‑model team, is an open‑source framework that lets creators generate video‑aligned sound effects while explicitly controlling content, style, and timing through text prompts, video dubbing, or reference audio, achieving SOTA performance on multiple benchmarks.
