Source record

Published by Hacker News · Show HN (50+ points) (T2 source). The original is at https://news.ycombinator.com/item?id=49026810.

Pipeline notes

The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.

Builder presents Echo, a system that routes each prompt across a pool of open-weight models, deciding compute allocation and combining outputs. Author reports it matches Fable's aggregate benchmark results at roughly one-third the inference cost and outperforms the best individual model in the pool.

Concrete first-hand evidence that per-prompt model routing and combining can cut inference cost meaningfully, with a usable chat interface and OpenAI-compatible API.

Cited by

No citations on record.