Verified reference stack
Cybersecurity — Security Log Analysis — RTX 4090
Replaced Qwen3-32B launch profile with Qwen3-14B to avoid marginal VRAM/context behavior in the reference launch stack. Qwen3-32B remains catalogued as an advanced/conditional option.
Ollama
Qwen3-14B
RTX 4090 24 GB / 64 GB RAM
EXCELLENTLocal-network + local-inference profile
Apache-2.0
Evidence
Freshness rule: Material changes set this stack to REVALIDATION_REQUIRED.