signal极客公园zh source2026-09-22
Jev，让全球程序员玩疯了
Former OpenAI researcher Diogo Almeida's Jev - a "System One" model that outputs only probability judgments, never text - sparked ~500 open-source projects within days, with costs of $0.042/million input tokens (output free) cutting monthly decision costs from ~$35K to ~$120 at 10K decisions/day. Applications include compressing a ~1M-token Claude Code context to 86K in one second, cleaning 9,081 records in 13 minutes for $0.32, and completing 9 Android-phone operations in 21 seconds via mobile-jev. The breakthrough splits "judgment" from "generation," proving cheap, millisecond-class probability models can handle high-volume infrastructure tasks while expensive generative LLMs are reserved only for human-facing text.
- why now
- Jev上线数天催生近500项目，成本骤降成新风口。
- evidence
- 6 sources carry this story · confidence high
- topic
- AI Tech & New Models
- source
- 极客公园
#Jev
score
score 9 out of 10. 0-10: how dense the facts are, multiplied by how much you can do with them after reading. 8+ means the topic's evidence bar is met: benchmarks and availability for a new model, amount and investors for a funding round, revenue figures for a solo-money story. Below 5 an item does not enter the digest. A press release scores 3 or less, a reprint loses 2, anything older than 14 days loses 1, a headline that misleads loses 3.
read the source