Classifying 1,018 AI Papers for $4: A Two-Model Pipeline at 256ms Median Latency
nutlope · x · 2026-09-17
Developer Nutlope built a pipeline to classify 1,018 AI research papers for about $4 total: DeepSeek V4 Flash summarized each paper ($3.99 on Together AI), then Jev classified each into one of 24 topics ($0.08 on Typesafe AI), with 256ms median end-to-end latency per paper. Results power a live paper-explorer site. His takeaway: use different models for different workflow stages instead of one model for everything; evals are running before replacing current classifications.
More from coding & agent
- Open weights aren't enough: four ownership tests for your personal AI memory and harness — sujingshen · 2026-09-17
- Asimov goes live every Friday to demo its AI coding agent in action — freelerobot · 2026-09-17
- Diorama gives OpenAI Codex coding agents a visual office you can watch work in real time — davidfromkansas · 2026-09-17
- Code-first, UI on top: building bespoke brand design tools with AI — floguo · 2026-09-17
- Study of 7 models across Claude Code, Codex, Pi: harness barely affects success but swings cost — DavideCrapis · 2026-09-17
- AI trading bot built with Jev is down 85%, owner shrugs it off — generativist · 2026-09-17