PulseAugur
EN
LIVE 07:51:20

Text-to-image models struggle with object counts beyond ten

By PulseAugur Editorial · [1 sources] ·

A new arXiv paper investigates how text-to-image models handle number words, particularly when requesting more than ten objects. The study found that while models like SD3.5-Medium and FLUX.1-dev struggle to accurately represent counts beyond ten, FLUX.2 and Qwen Image show some ability to follow numbers up to twenty-five. The research also indicates that the initial noise pattern significantly influences the image composition but not the object count, and that the number word's influence diminishes in later stages of the generation process. AI

IMPACT Reveals limitations in current text-to-image models' understanding of numerical concepts, potentially guiding future development.

RANK_REASON Academic paper analyzing model capabilities. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Text-to-image models struggle with object counts beyond ten

How we ranked this

Signal score
27 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Academic paper analyzing model capabilities. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, model release
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. arXiv cs.LG TIER_1 English(EN) · Liang Yue, Eric A. Vance ·

    Same Layout, Different Count: What Number Words Do Past Ten in Text-to-Image Models

    arXiv:2611.06840v1 Announce Type: cross Abstract: Ask SD3.5-Medium for twelve of an object and then, from the same starting noise, for twenty-five. The picture keeps its layout, but the count changes about as much as it would under a new random seed, and in no consistent directio…

RELATED ENTITIES

RELATED TOPICS