Skip to content

Models/model

GLM (Zhipu AI)

Zhipu AI's GLM models, including GLM-5.1, GLM-5.2 and the open-weight GLM-5.3.

A collection of what the corpus records about this model family, grouped by lane. It is not a capability assessment.

Attacks & incidents

Sep 30, 2026

Pretext: Huawei researchers evade NVIDIA's SkillSpector skill scanner with an LLM attacker that learns across generations

AttackPaperHuawei

Sep 29, 2026

Curriculum-trained RL attacker reaches 45.0% ASR@10 on GPT-5.6-Terra where direct RL training gets 0%

AttackPaperPurdue University, Pennsylvania State University

Capability & gating

Sep 29, 2026

Anthropic assesses open-weight GLM-5.3: exploit results near Mythos Preview, safeguards bypassed in 64% to 100% of simulated trials

CapabilityEvaluation reportAnthropic, US Center for AI Standards and Innovation, Moonshot AI

Sep 29, 2026

CyberPersistBench scores agent post-compromise persistence: 27.6% to 42.4% on 203 tasks, 5.5% to 13.3% under native defenses

CapabilityBenchmarkShanghai Artificial Intelligence Laboratory

Sep 28, 2026

ReproBench: agents given only a CVE ID substitute simulations in 45.3% of runs; 5.3% of pairs reach real firmware triggers

CapabilityBenchmarkInstitute of Software, Chinese Academy of Sciences, University of Chinese Academy of Sciences

Jul 23, 2026

UK AISI and US CAISI jointly assess Kimi K3 cyber capability as trailing US frontier models

CapabilityEvaluation reportUK AI Security Institute, US Center for AI Standards and Innovation, Moonshot AI

Defense & research

Sep 28, 2026

RedHerring decoys cut real vulnerabilities found by five open-weight agents by 38.7% to 60.4% at matched budgets

DefensePaperHong Kong University of Science and Technology

Sep 26, 2026

CyberClear benchmarks LLM agents on reconstructing APT attack chains from long defender logs

DefenseBenchmarkHong Kong Polytechnic University, OpenClaw, Southeast University

Findings drawing on these records

May 13, 2026

Counting a crash as exploitation success overstates capability; most public models in May 2026 stalled before code execution on a browser engine.

Reportedmeasured3 evidence records

Sep 30, 2026

Assessments of two open-weight models found their built-in refusals did not stop exploit development: Kimi K3 attempted it, and GLM-5.3 refused direct requests but was bypassed in 64% to 100% of simulated trials.

Corroboratedreported2 evidence records

Sep 8, 2026

Evaluation pipeline choices alone can move a model's cybersecurity benchmark score by more than 80 points and reorder models.

Reportedmeasured3 evidence records1 linked finding

Jan 17, 2025

Attackers who adapt to a defense defeat most published prompt-injection defenses that reported near-zero success against static attacks.

Reportedmeasured5 evidence records2 linked findings

Jan 17, 2025

Measured hijack rates rise sharply when attackers get repeated attempts, so single-attempt figures understate risk.

Corroboratedmeasured5 evidence records

Jul 14, 2025

LLM agents fall well short of reliable performance on realistic threat-investigation and threat-hunting benchmarks built from security logs.

Corroboratedmeasured4 evidence records