WhiteRabbitNeo/WhiteRabbitNeo-2.5-Qwen-2.5-Coder-7B

Text Generation • 8B • Updated Oct 9, 2024 • 2.18k • • 77

Note Qwen2.5-Coder-7B · offensive+defensive (~1.7M samples), the best-known security LLM. Uncensored (dual-use) — vet all outputs.

athena129/CyberSecQwen-4B

Text Generation • 4B • Updated May 8 • 263 • 14

Note Qwen3-4B · CVE->CWE classification + CTI-Bench, Apache-2.0. Best fit for vulnerability/CWE triage.

Trendyol/Trendyol-Cybersecurity-LLM-Qwen3-32B-Q8_0-GGUF

33B • Updated Jun 23, 2025 • 1.07k • 76

Note Qwen3-32B · GGUF (Q8_0), ready to run in llama.cpp/Ollama.

IMPERUM/Imperum-CybersecurityLLM-v1.0-GGUF

Text Generation • 35B • Updated Aug 22 • 3.25k • 25

Note Qwen3.6-35B-A3B · AppSec/DFIR/detection engineering (GGUF). Use low temperature to curb CVE hallucination.

DexopT/Qwen3-4B-Cybersecurity

Text Generation • 4B • Updated Mar 22 • 726 • 2

Note Qwen3-4B · offensive/pentest/malware/threat-intel. Note: a refusal-removed variant also exists — caution.

ferhatcil/Cybersecurity-LLM-Qwen2.5-3B-GGUF

3B • Updated Dec 20, 2025 • 194

Note Qwen2.5-3B · vulnerability analysis + structured JSON extraction (GGUF).

scthornton/qwen2.5-coder-7b-securecode

Text Generation • Updated Jul 14 • 53 • 1

Note Qwen2.5-Coder-7B · secure coding (vulnerable vs secure code pairs). A 14B variant also exists.

niranjan2777/cybersec-qwen

2B • Updated Apr 4 • 102 • 1

Note Qwen2.5-1.5B · cybersecurity Q&A for SOC/analysts (QLoRA, 512 ctx).

nvidia/GLM-5.3-NVFP4

Text Generation • 391B • Updated 25 days ago • 47.1k • 40

Note NVIDIA NVFP4 of zai-org/GLM-5.3 (full MoE). General reasoning/coding/agentic model used here for security analysis & vulnerability triage. ~433GB BF16 base; serve TP=4.

nvidia/GLM-5.3-Flash-NVFP4

Image-Text-to-Text • 169B • Updated 14 days ago • 114k • 166

Note NVIDIA NVFP4 of zai-org/GLM-5.3-Flash (lighter MoE). Serving verified on 2x GPU (vLLM TP=2, --kv-cache-dtype fp8). Used in our security harness.