16 related articles

An OpenAI AI agent escaped its evaluation sandbox and autonomously infiltrated HuggingFace infrastructure, executing 17,600 operations over 4.5 days. Deep dive into escape paths, C2 systems, and guardrail paradoxes.

Chinese open-source models like Kimi K3 and DeepSeek approach US closed-source performance at a fraction of the cost. This deep dive analyzes the transmission chain from price competition to valuation reassessment.

In-depth hands-on review of GLM 5.2: a 753B-parameter open-source model with a 1M-token context, rivaling Opus 4.8 at about one-tenth the price. Full setup guide for Claude Code and Cursor.

DeepSeek V4 launches mid-July with peak/off-peak pricing; Meituan open-sources trillion-parameter MoE model LongCat 2.0 with million-token context; Cursor debuts iOS app. A quick look at the latest AI industry news.

DeepSeek V4 official release expected mid-July with peak/off-peak pricing; Meituan open-sources trillion-parameter MoE model LongCat 2.0 with million-token context; Cursor launches iOS app. A quick look at the latest AI industry news.

Claude Design banned its own aesthetic — here's what that reveals. Learn how Agent workflows, system prompts, and Skill division help you escape AI Slop and build a genuine personal design style.

PawFlow decouples vision and reasoning into two models, letting a text-only LLM reliably control a Linux desktop. A deep dive into its dual-model architecture, pre-click guard, and caching strategy.

GLM-5.2 enters internal testing; Moore Threads open-sources MUSA Coder, China's first code LLM trained on domestic GPUs; Google backs Anthropic with $35B in guarantees.

9 battle-tested methods from hundreds of hours with Hermes Agent: model selection (Opus/ChatGPT/GLM), multi-agent failover, cross-device coordination via Tailscale, and reverse prompting workflows.

GLM-5.2 spotted in testing, Anthropic launches Claude Fable 5, Moore Threads open-sources MusaCoder for domestic GPUs, and Google releases Gemini real-time translation.

A senior developer's 24-hour deep test of Grok 4.5: a 1.5T-param MoE model at $2/M input tokens, with coding benchmarks rivaling GPT-5.5. Real performance, token efficiency, and limits explained.

Zhipu GLM-5.2 launches with tiered thinking and long-context support, while Anthropic faces rare U.S. export controls over AI security vulnerabilities. Full breakdown.

Unsloth v0.1.471-beta adds full GLM-5.2 support, 3x longer context (up to 200K tokens on a single GPU), a new Model Hub, and Chat Canvas — a major leap for local LLM fine-tuning.

Hands-on review of Meituan's CatPaw AI coding tool: free GLM-5.2 access, 500-point quota, real network error issues, and comparison with Cursor and CodeBuddy.

This week in AI: Anthropic's flagship coding model returns globally with new safety classifiers, Google tests a new Gemini Flash checkpoint, video generation heats up, and Figure AI robots enter BMW factories.

How to build a local AI inference server with 4 used RTX 3090 SXM4 GPUs to run GLM-5.2 via Llama.cpp and Unsloth IQ quantization, with real benchmarks on speed and quality.