Will DePue pitches OpenAI: add API-style fan-out so multi-agent systems dispatch 1000s of tiny cheap models
willdepue · x · 2026-10-28
- Will DePue asks OpenAI researchers (Polynoamial, Max Schwarzer et al.) to add a Jev/Luna-like API model capability for multi-agent systems.
- His workflow: when doing fast data analysis, he programmatically fans out cheap model calls — e.g. have the model write a Python script that dispatches 1000 Lunas to classify each message, then dig into only the flagged data; or run Jev/Luna on every file to catch bugs, at the cost of one Sol subagent.
- His point: models aren't trained for API-style delegation, so today's multi-agent setups are only efficient for small splits, not fanning out into hundreds or thousands of short calls.
More from coding & agent
- ZCode offers 300M free GLM-5.3-Flash tokens daily, with a quirky claiming window — AccBalanced · 2026-10-28
- Dev ships 3-month project in 2.5 weeks by letting Codex self-verify in a loop — kmeanskaran · 2026-10-28
- MLE-World: a learned world model predicts ML experiment outcomes to cut agent costs — Brian Ko · 2026-10-28
- Marc Andreessen on Replit: the dawn of the 'self-driving company' — amasad · 2026-10-28
- Accent-induced bug: speech-to-text turns "Claude Instance" into "cloud instance" for a coding agent — francesc · 2026-10-28
- Harrison Chase on self-improving agents: LangChain aims to be the model-agnostic all-in-one platform — hwchase17 · 2026-10-28