AI Paper Atlas

Research map / Large language models / Agents, code and math

AetherCode: Evaluating LLMs’ Ability to Win In Premier Programming Competitions

Zihan Wang, Jiaze Chen, Zhicheng Liu, Haojie Pan, Markus Mak, Yidi Du, Geonsik Moon, Aaron Tua, Kunshuo Peng, Jiayi Lu, Boqian Zou, Chenyang Ran and 14 more

ICLR 2026 · Large language models › Agents, code and math · approach: Mathematical and structural reasoning benchmarks

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use reasoning models instead of non-reasoning modelsLLM code generation · AetherCode v1Pass@1 (%): o4-mini-high 35.5 vs GPT-4.1 10.5Improved
sample more attempts (Pass@4 vs Pass@1) on top-tier modelso4-mini-high · AetherCode v1Pass@k (%): 35.5 -> 46.6Improved
sample more attempts (Pass@4 vs Pass@1) on a weaker modelQwen3-32B · AetherCode v1Pass@k (%): 16.3 -> 23.9Improved
sample four attempts for non-reasoning modelsnon-reasoning models · AetherCode v1Pass@4 (%): GPT-4.1 10.5 -> 15.3No clear effect
add expert-annotated hack test cases to automatically generated test casesAetherCode test suite · AetherCode test casesTNR: automatic phase alone could not reach 100% TNR; hybrid reaches 100%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy