All resources / AI Models & Benchmarks
FreeResearch Paper
AI Models & Benchmarks
The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMs
How to cite it
Built from the details listed on this page. Check them against the paper before you submit.
APA
(n.d.). The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMs. https://arxiv.org/pdf/2509.09677
MLA
"The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMs." n.d., https://arxiv.org/pdf/2509.09677.
BibTeX
@article{paperndthe,
title = {The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMs},
year = {n.d.},
url = {https://arxiv.org/pdf/2509.09677},
}What it is
arXiv paper (September 2025) arguing that LLM accuracy on long tasks is limited by per-step error compounding rather than reasoning ability, and introducing self-conditioning as a key failure mode when models see their own past outputs.
Added Oct 16, 2026 · 0 opens