The LGTM (Page 100)

The LGTM

Sign in Subscribe

Microsoft’s MDASH Says the Security Agent Is a System, Not a Model

Microsoft’s MDASH Says the Security Agent Is a System, Not a Model

Microsoft’s MDASH announcement looks like another frontier-AI benchmark post until you read the architecture. Then the actual story snaps into focus: the security agent is not a model. It is a factory. Microsoft says its new multi-model agentic security harness helped find 16 Windows vulnerabilities in the May 12

Frontier Cyber Models Are Now Outrunning the Benchmarks Built to Measure Them

Frontier Cyber Models Are Now Outrunning the Benchmarks Built to Measure Them

The scary part of the UK AI Security Institute’s latest cyber-model update is not that Claude Mythos Preview or GPT-5.5 can “do security work.” We already knew that. The scary part is that the measurement apparatus is starting to look underpowered. AISI says frontier models have now substantially

Unsloth’s Qwen3.6 MTP GGUFs Make Local Coding Agents Less Theoretical

Unsloth’s Qwen3.6 MTP GGUFs Make Local Coding Agents Less Theoretical

Unsloth’s new Qwen3.6-27B MTP GGUF release is not interesting because the internet needed another quantized model file. It is interesting because it documents the messy runtime details that decide whether local coding agents feel usable or collapse into a pile of parser errors, exhausted token budgets, and “works