# PORA lets LLM agents decide when to think longer

> The new PORA method teaches language model agents to pick short or long reasoning based on budget and expected benefit, cutting compute by up to half.

Oossa · 2026-10-21 · https://oossa.com/en/pora-lets-llm-agents-decide-when-to-think-longer

Researchers led by Qi Zhang released PORA, a technique for LLM agents that learns when longer reasoning improves results. During training the system runs the same step twice—once with a short reasoning limit and once with a longer one—to see which yields a better final outcome. At run time a controller uses that learned estimate, plus the remaining computation budget, to choose the cheaper option without running both. Tests on memory, code, retail and math tasks showed up to 32.9% higher scores on a memory benchmark and up to 31% fewer model calls on math tasks.

## The facts

- Paper posted on arXiv on 2026-10-21
- On BFCL memory benchmark PORA outperforms the prior best by 17.4–32.9 percentage points

## Why it matters

It lets developers run smarter AI agents that use less cloud compute while keeping performance, lowering costs for services that rely on multi‑step reasoning.

## Sources & references

1. [PORA: Learning When More Reasoning Pays Off for LLM Agents](https://arxiv.org/abs/2610.27119) – arXiv, 2026-10-21

Last updated: 2026-10-21