JEV vs local cross-encoder as RAG reranker: same accuracy, worse economics
ExamImmediate8956 · reddit · 2026-09-28
Testing TypeSafe JEV as a RAG reranker on the 144-page EU AI Act (937 chunks, 18 answerable + 3 unanswerable questions), all three setups hit 17/18 accuracy: no reranker (10,278 tokens, 8s), local cross-encoder (626 tokens, 3.2s, free), JEV (476 tokens, 3.9s, $0.0018/query). JEV returned empty candidate sets on unanswerable questions, saving only $0.000074/query — far less than its own cost. The takeaway: benchmark baselines matter; JEV looks great against "send everything to the LLM" but loses to a free local reranker.
More from coding & agent
- Harmony Launches 100+ AI Agents in Slack for IT and HR Automation — Codeblix_Ltd · 2026-09-28
- Open-source Editable Design: real text, independent layers, no more pixel-locked AI graphics — jiqizhixin · 2026-09-28
- INT21's agent-built factory ships 20 Rust inference engines across 7 model types — bingxu_ · 2026-09-28
- pi-demo plugin lets the Pi agent record, edit, and ship screen-capture reels — solyarisoftware · 2026-09-28
- Dev Built AI Agents That Find a Business Idea and Run the Whole Operation — Ok-Werewolf-7293 · 2026-09-28
- Devin user at $100k+ run-rate questions if coding agent revenues will stick — brandon_galang · 2026-09-28