Source record

Published by Hacker News · Show HN (50+ points) (T2 source). The original is at https://github.com/experientiallabs/world-model-optimizer.

Pipeline notes

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

Open-source tool 'world-model-optimizer' that trains LLM routers for agent workloads by simulating production tool responses through text world modeling, using KNN-based model selection across frontier, open-source, and local models. Supports cost or quality optimization, confidence gating, and cache-aware pricing via three CLI commands (build, optimize, serve).

Open-source agent router with specific CLI workflow and benchmark results showing 44-66% cost reduction across RouterBench, TauBench, and Terminal Bench 2, directly reusable for multi-model agent stacks.

Cited by

No citations on record.