AI Core

AREX-2 defines agent self-improvement as turning more test-time rounds into a better solution, synthesizes long-horizon improvement trajectories from machine learning engineering and algorithmic programming tasks that retain failures and regressions, and after fine-tuning Qwen3.8-27B reaches 81.8 on MLE-bench Lite and 70.7 on Frontier-CS while lifting BrowseComp to 84.0, HLE to 52.6, GAIA to 92.2 and DeepSearchQA to 93.8 without adding new deep-research data, with continued gains as the round budget grows.

The work factorizes masked diffusion language model (MDM) inference into five axes—score, cardinality, region, commitment, and planning—defines adaptation opportunity as the one-step utility advantage of the best candidate action over a validation-selected fixed action, finds across three models (LLaDA-8B-Instruct, LLaDA-1.5, Dream-7B) and ten tasks that adaptation opportunities are highly heterogeneous, and proposes selective adaptation in which lightweight detectors calibrated on validation prompts identify high-opportunity states, capturing for example 56.9% of the candidate-set oracle opportunity by adapting only the top 10% of states on LLaDA-8B constrained JSON filling, with end-to-end selective decoding raising task-macro utility from 0.392 to 0.424 (LLaDA-8B), 0.344 to 0.

PatchHolmes is a two-phase patch retrieval system whose first stage fuses BM25 with time decay and Qwen3-Embedding-8B dense retrieval via RRF into a top-100 candidate set, and whose second stage runs a frozen open-weight LLM agent that views all 100 candidates listwise through four tools (list_candidates, read_commit, read_file_diff, submit_answer) and submits a single best commit; on 809 CVEs from GitHubAD it reaches 59.95% Recall@1, 25.34% above the pointwise classifier Favia and 31.40% above IRCoT, adds 27.32% Recall@1 over the retriever's top pick on identical candidates, and transfers unchanged to PatchFinder_top10 to lift Recall@1 from 24.28% to 39.86%.

LEAP splits an hour-scale recording into fixed-duration blocks, runs a lightweight localization pass per block to score short candidate windows, then re-encodes only the top-ranked windows in a single bounded answer pass, so the answer input and peak context stay independent of recording duration; across several AVQA benchmarks it improves over the Qwen3-Omni-30B-A3B baseline by 4.5–16.8% and transfers to MiniCPM-o 4.5, surpassing its published results by 3.1–13.0%.

The work introduces EvoDuet, a bi-level method that co-evolves solutions and web search queries with fixed model parameters: at each iteration a knowledge-gap-based retrieval gate decides whether to retrieve new documents, reuse stored ones, or proceed without them, an inner loop refines queries and ranks documents by the solution scores they are predicted to yield, and an outer loop generates candidates in parallel and records evaluated outcomes for later searches; across 21 optimization tasks with one candidate per iteration it raises OpenEvolve's normalized discovery gain from 74.1% to 78.0% with GPT-5.6-Luna and from 61.3% to 82.3% with Gemini-3.8-Flash, Qwen3.

The work shows that most of the reported gain from jointly decoding all words in a sentence (d'Ascoli et al., 2025) is reproducible on synthetic signals containing no brain information (22.0% versus 22.3% on real MEG), because neighbouring three-second word-aligned windows overlap and implicitly leak word duration; decoding words independently instead (SimpleB2T) makes aggregation across observations and an LLM prior effective, reaching 36.6% word error rate with five observations per word on a clinically motivated benchmark.

The authors propose RegLLM, a diagnostic harness for bounded autonomy in regulated agentic workflows that instruments six trustworthiness signals (citation validity, source grounding, schema compliance, escalation correctness, constitutional alignment, unsafe-action rate) alongside a deterministic runtime supervisor; an offline reference run (n=12) lifts escalation recall from 0 to 0.67 and cuts the unsafe-action rate from 0.33 to 0.08, while two nominally identical single-GPU Qwen2.5-3B LoRA/DPO pilots (n=8, same seed and eval split) show task success of 0.25 versus 0.12 and escalation recall of 1.0 versus 0.

This work systematically evaluates GPT-6 Astra as an embodied policy across six domains—gripper manipulation, dexterous manipulation, mobile manipulation, navigation, locomotion, and humanoid loco-manipulation—comparing direct control with hybrid control that cooperates with learned policies or whole-body controllers, finding leading navigation results (92% on RxR instruction following, 82% on HM3D object search), hybrid success of 48% on RoboDojo, 50% on DexJoCo, and 38.7% on RoboCasa365, while direct in-hand control and dense-reference locomotion remain unreliable, with substantial token use and inference latency recorded.

The work introduces the Agentic Idea Manager (AIM), a fully autonomous framework that makes research ideas explicit search objects: a Bayesian-optimization-inspired Agentic Surrogate organizes ideas into semantic clusters and produces ordinal promisingness estimates, an Agentic Acquisition mechanism balances exploration and exploitation at cluster and idea level, a Solution Auditor checks idea-solution integrity, and a Resource Planner adaptively allocates parallel branches under a fixed budget; on 10 AutoLab tasks AIM reaches 67.0% average on System Optimization and 55.8% on long-horizon Model Development & CUDA, exceeding the strongest baseline ScientistOne by 1.6 and 4.9 percentage points and reaching that baseline's best score up to 3.1x faster in wall-clock time.