Retrieval Quality at Context Limit

TL;DR

This paper investigates how large language models perform in retrieving information from long contexts, revealing that Gemini 2.5 Flash maintains high accuracy even at context limits, unlike previous models affected by the 'Lost in the Middle' phenomenon.

Contribution

It demonstrates that Gemini 2.5 Flash effectively overcomes the 'Lost in the Middle' issue, showing robust retrieval capabilities at context limits for factoid questions.

Findings

Gemini 2.5 Flash answers needle-in-a-haystack questions accurately regardless of document position.

The 'Lost in the Middle' effect is not observed in Gemini 2.5 Flash for simple factoid Q&A.

Substantial improvements in long-context retrieval are achieved with Gemini 2.5 Flash.

Abstract

The ability of large language models (LLMs) to recall and retrieve information from long contexts is critical for many real-world applications. Prior work (Liu et al., 2023) reported that LLMs suffer significant drops in retrieval accuracy for facts placed in the middle of large contexts, an effect known as "Lost in the Middle" (LITM). We find the model Gemini 2.5 Flash can answer needle-in-a-haystack questions with great accuracy regardless of document position including when the document is nearly at the input context limit. Our results suggest that the "Lost in the Middle" effect is not present for simple factoid Q\&A in Gemini 2.5 Flash, indicating substantial improvements in long-context retrieval.