InfoResearchPreprintLLM-specific
JEV vs. LLMs: A Comparative Analysis for Phishing Email Detection
- Published
- Record updated
Summary
This paper compares Jev 1.13, a typed decision model from TypeSafe AI, with Gemini 3.5 Flash-Lite, NVIDIA Nemotron 3 Ultra, and Meta Llama 4 Maverick on 1500 labelled phishing emails. Jev reports the highest accuracy (79.67%), precision (71.43%), recall (97.97%), and F1 score (82.62%), along with the shortest batch time (173.88 seconds), lowest token count (597,895), lowest cost (USD 0.015103 per 1000 decisions), and lowest Brier score (0.1698) and ECE (0.1594). The authors state that all four systems require further validation before standalone deployment.
Related items
- InfoCramer says Nvidia’s woes are trumping Anthropic excitement for the rest of the AI tradeSame vendor · CNBC Technology
- InfoA brief history of Nvidia’s recent stock fluctuationsSame vendor · The Verge (AI)
- InfoMeta’s Muse Charm might cost $300Same vendor · The Verge (AI)
- InfoGoogle is putting AI chats right on the homepageSame vendor · The Verge (AI)
- InfoThreatsDay: GPT-6.2 Jailbreak Flaws, BubbleLoader PNG Malware, $500M Ransom Demands + 17 More StoriesSame vendor · The Hacker News