---
title: "Apple researchers cut multimodal agent latency 90 per cent with SSR"
description: "A new framework from Apple Machine Learning Research trades open-ended generation for selection, slashing reasoning time while keeping success rates competitive."
url: https://mactoreality.com/apple-researchers-cut-multimodal-agent-latency-90-per-cent-with-ssr
published: 2026-10-16T06:42:26Z
updated: 2026-10-16T06:42:26Z
author: "Mac to Reality Newsdesk"
category: News
tags: ["Software"]
publisher: Mac to Reality
ai_assisted: true
apple_heat_impact: 1
---

# Apple researchers cut multimodal agent latency 90 per cent with SSR

*A new framework from Apple Machine Learning Research trades open-ended generation for selection, slashing reasoning time while keeping success rates competitive.*

**Key points**

- Apple ML Research introduced Selection-Based Structured Reasoning (SSR), a framework for efficient multimodal search agents.
- SSR reduces per-turn reasoning latency by over 90 per cent and total pre-action latency by 28–54 per cent.
- The framework achieves success rates competitive with leading search agents of the same scale on multimodal benchmarks.

[Apple Machine Learning Research published a paper on 15 October](https://machinelearning.apple.com/research/selection-structured-reasoning) introducing Selection-Based Structured Reasoning (SSR), a framework designed to cut reasoning latency in multimodal search agents. Instead of generating open-ended reasoning before every action, SSR treats reasoning as a selection problem, drawing from a set of pre-specified, reusable natural-language candidates that can be scored in parallel.

The authors, a team from Apple and the University of Illinois Urbana-Champaign led by Feiyu Gavin Zhu, Xiaoyu Zhu and colleagues, evaluated SSR on seven multimodal search benchmarks using models with 2 billion and 4 billion parameters. The results show the framework matches the success rate of leading search agents at the same scale while cutting per-turn reasoning latency by more than 90 per cent. Total per-query model inference latency fell by 28–54 per cent. (The researchers’ paper does not name the specific benchmarks or the agents it was tested against.)

The work builds on Apple’s earlier research into efficient reasoning, including Interleaved Reasoning for LLMs via Reinforcement Learning and Learning Structured Reasoning via Tractable Trajectory Control.

## Why it matters

Small models that produce long, freeform reasoning waste time and compute for little gain. SSR’s shift to a selection-based design lets a modest 2B or 4B-parameter model act decisively. [Apple researchers have been pursuing similar efficiency gains](https://mactoreality.com/apple-researchers-propose-selfcompact-to-let-ai-agents-self-summarise-reasoning-context) in other agent contexts. In our view, SSR is a practical step towards running capable multimodal agents locally, where latency and compute are strict constraints.

## Sources

- [Apple Machine Learning Research: Selection-Based Structured Reasoning: Toward Efficient Multimodal Search Agents](https://machinelearning.apple.com/research/selection-structured-reasoning)

*This report was written by the Mac to Reality newsdesk with AI assistance from the sources listed above, under the editorial policy at https://mactoreality.com/editorial-policy.*