AREX-2
An LLM agent framework with self-improvement capability that iteratively refines solutions at test time through two domain-agnostic mechanisms—reflection and long-horizon execution—while supporting cross-domain capability transfer.
Timeline (last 90 days)
The core objective of AREX-2 is to advance the self-improvement capability of LLM agents to iteratively optimize their own solutions at test time
AREX-2 decomposes self-improvement capability into two complementary mechanisms: reflection and long-horizon execution
AREX-2 is based on a core assumption that reflection and long-horizon execution capabilities are domain-agnostic and can be transferred from scenarios suitable for supervised training to different task domains
AREX-2 synthesizes long-horizon improvement trajectories from two domains—machine learning tasks and algorithmic programming tasks—as training data
AREX-2 is built on Qwen3.8-27B
AREX-2 scored 81.8 on MLE-bench Lite and 70.7 on Frontier-CS
AREX-2 achieved the following scores on deep research tasks: BrowseComp 84.0, HLE 52.6, GAIA 92.2, and DeepSearchQA 93.8
Despite training data coming from machine learning and algorithmic programming, AREX-2 can transfer to deep research tasks, supporting the assumption that capabilities are domain-agnostic
As the iteration budget increases, the AREX-2 agent continues to improve rather than hitting a ceiling prematurely
Evaluation metrics for machine learning tasks and test cases for algorithmic programming tasks provide objective, automated feedback, making large-scale trajectory synthesis possible and avoiding reliance on manual annotation
1 more timeline events
All Facts (11)
The core objective of AREX-2 is to advance the self-improvement capability of LLM agents to iteratively optimize their own solutions at test time
50%UnverifiedAREX-2 decomposes self-improvement capability into two complementary mechanisms: reflection and long-horizon execution
50%UnverifiedAREX-2 is based on a core assumption that reflection and long-horizon execution capabilities are domain-agnostic and can be transferred from scenarios suitable for supervised training to different task domains
50%UnverifiedAREX-2 synthesizes long-horizon improvement trajectories from two domains—machine learning tasks and algorithmic programming tasks—as training data
50%UnverifiedAREX-2 is built on Qwen3.8-27B
50%UnverifiedAREX-2 scored 81.8 on MLE-bench Lite and 70.7 on Frontier-CS
50%UnverifiedAREX-2 achieved the following scores on deep research tasks: BrowseComp 84.0, HLE 52.6, GAIA 92.2, and DeepSearchQA 93.8
50%UnverifiedDespite training data coming from machine learning and algorithmic programming, AREX-2 can transfer to deep research tasks, supporting the assumption that capabilities are domain-agnostic
50%UnverifiedAs the iteration budget increases, the AREX-2 agent continues to improve rather than hitting a ceiling prematurely
50%UnverifiedEvaluation metrics for machine learning tasks and test cases for algorithmic programming tasks provide objective, automated feedback, making large-scale trajectory synthesis possible and avoiding reliance on manual annotation
50%UnverifiedThe core conclusion of the AREX-2 paper is that long-horizon reflection data is an effective path toward self-improving agents
50%