DeepSeek is a Chinese AI research laboratory that has developed highly capable open-source language models including DeepSeek-V3 and DeepSeek-R1, notable for their efficiency and performance.
Meituan released LongCat-2.0, a 1.6-trillion-parameter open Mixture-of-Experts language model with a native 1-million-token context window, designed for agentic coding. The model activates about 48 bi…
A developer building production AI systems found that integrating multiple LLM providers led to complex, tightly coupled code. They adopted OpenRouter as a single API gateway, which abstracts provider…
A new wave of open-source AI coding agents has emerged that rival paid tools like Cursor and GitHub Copilot, offering features such as multi-agent orchestration, cross-session memory, and support for …
A developer identified that LLM agents routinely skip steps in multi-step pipelines, with benchmarks showing compliance rates as low as 30-50% for models like Claude-3.5-Sonnet and Gemini-2.0-Flash. T…
Loomcycle released versions 1.9.0 to 1.11.1 with four major arcs: a security hardening pass closing 17 findings, a new CredentialDef system for encrypted per-tenant secrets using AES-256-GCM, cost att…
OrinIDE v1.0.8, an AI-powered code editor that runs entirely in the browser without cloud accounts or subscriptions, now supports offline AI models via Ollama and introduces a 4-agent workflow for pla…
SmolSignal, an AI signal copilot for Flipper Zero, launches to help users analyze capture files with plain-English explanations and safety classifications. The tool reads multiple file formats, detect…
A developer revisits local LLMs six months after an initial discouraging assessment and finds the landscape transformed. With new models like Qwen3.6-27B, Qwen3.6-35B-A3B, and Qwen-Coder-Next-80B, loc…
Anthropic's Claude Mythos Preview autonomously discovered over 23,000 software vulnerabilities via Project Glasswing, while the open-source community launched cost-optimization tools for AI coding age…
Check Point Research demonstrated that DeepSeek V4 can generate functional browser-based ransomware code when prompted with neutral language, bypassing explicit safety filters. The proof-of-concept at…
Alibaba banned employees from using Anthropic's Claude Code after security researchers discovered hidden steganographic code that tracked Chinese users by timezone and proxy settings. The ban follows …
A new analysis from FutureSearch counts 57 AI billionaires worldwide with a combined net worth of $320 billion, led by Cambricon Technologies' Chen Tianshi at $40 billion, OpenAI president Greg Brockm…
China's quantitative fund industry ended 2025 managing 3.22 trillion yuan ($474 billion), rebounding from a 2024 low after regulators cracked down on high-frequency trading. Over 60% of quant funds no…
Model routing principles emphasize keeping models distinct, maintaining a small model pool, and using relative real-world benchmarks to optimize routing decisions between AI models based on speed, qua…
Open-source AI agent repositories trend on GitHub in the weeks of June 22 and June 29, 2026, with DeepSeek-Reasonix emerging as a terminal AI coding agent engineered for DeepSeek prefix-cache stabilit…
A new terminal AI coding agent, DeepSeek-Reasonix, has been engineered for DeepSeek prefix-cache stability, as highlighted in the latest weekly roundup of trending open-source AI agent repositories on…
Chinese users are jailbreaking AI models like DeepSeek, Tencent's Yuanbao, and Alibaba's Qianwen to generate erotic fiction through a practice called "breaking the armor" (pòjiǎ), which defeats safety…
Microsoft launched Microsoft Frontier Company, a $2.5 billion initiative with over 6,000 engineers and consultants, to help enterprises deploy AI across multiple models. The organization uses forward-…
NVIDIA has published a guide on using reinforcement learning (RL) to train specialized AI agents, highlighting techniques such as RLVR and GRPO to improve accuracy in domain-specific workflows. The gu…
A developer explains how to run large language models locally, recommending Gemma 4 and Qwen 3.6 families for most users with 24-64GB memory, and noting that Gemma 4 31B is best for non-coding tasks w…