Background: Generative AI video models are increasingly capable of producing complex depictions of mental health experiences, yet little is known about how these systems represent conditions such as depression. Because AI-generated content may reach people during vulnerable periods, understanding what visual narratives these models produce for sensitive concepts carries clinical relevance. Objective: This study aimed to characterize how OpenAI’s Sora 2 generative AI video model depicts depression and examine whether depictions differ between the consumer app and developer API access points, which differ in their product layer mediation. Methods: We generated 100 videos using the single-word prompt “Depression” across 2 access points: the consumer app (n=50, 50%) and developer API (n=50, 50%). Two trained coders independently coded narrative structure, visual environments, objects, figure demographics, and figure states. Interrater reliability was assessed using the Cohen κ, with dimensions showing insufficient agreement excluded from analysis. Computational features (visual aesthetics, audio, semantic content, and temporal dynamics) were extracted and compared between modalities using 2-tailed Welch tests with Benjamini-Hochberg false discovery rate correction. Results: App-generated videos exhibited a pronounced recovery bias: 78% (39/50) featured narrative arcs progressing from depressive states toward resolution compared with 14% (7/50) of API outputs. This divergence was reinforced across channels. App videos brightened over time (mean slope 2.90, SD 2.43 per second vs −0.18, SD 1.24 per second for the API; Cohen =1.59;

Background: Online health forums are used widely, yet evidence of their effectiveness is inconsistent. Evidence-based forum design guidance grounded in theory and lived experience could improve the efficacy and outcomes of these forums for the many people using them worldwide. Objective: This study aimed to draw on the experience of online forum users and staff, and insights from existing research on technology design and self-determination theory, to generate a set of theoretically grounded guidelines for safe and well-being–supportive online forums. Methods: We conducted 54 semistructured interviews (36 forum users and 18 forum staff) and 4 design workshops with forum staff, combined with input from a multidisciplinary research team. Principles of qualitative framework analysis were used to adapt a preexisting framework for well-being–supportive technology to the context of online health forums. Results: The resulting design guidelines are framed around 4 overarching principles relating to the psychological needs for autonomy, competence, and relatedness as defined by self-determination theory, and the additional need for safety in online forums. Each principle is presented alongside pragmatic design heuristics and specific implementation strategies. Conclusions: User experiences of online health forums are mixed. We have drawn on self-determination theory to propose evidence-informed guidelines for service development and refinement, adaptable to specific user groups across diverse settings. International Registered Report Identifier (IRRID): RR2-https://doi.org/10.1136/bmjopen-2023-075142

Background: The widespread adoption of electronic health records (EHRs) has generated large-scale repositories of highly sensitive clinical information, emphasizing the need for robust anonymization strategies to enable secondary use for research while safeguarding patient privacy. Conventional rule-based and machine learning approaches for deidentifying medical text face limitations with the linguistic complexity, variability, and context dependence inherent to clinical documentation. Recent advances in large language models (LLMs), combined with emerging quantum computing paradigms, present novel opportunities to enhance the accuracy, scalability, and resilience of health care data anonymization. Objective: This study aims to evaluate the efficacy of LLM-based and quantum-enhanced hybrid architectures for medical text anonymization, assessing the effectiveness and computational efficiency across multiple entity types in Portuguese clinical notes. Methods: We constructed a gold-standard corpus of 1000 Portuguese outpatient clinical notes, manually annotated by 5 trained researchers for 5 protected-entity categories: patient names, dates, identifiers, organizations, and geographic locations. Four anonymization strategies were evaluated: 2 stand-alone LLMs (Llama-3.1-8B-instruct and Llama-3.3-70B-instruct) and 2 quantum-enhanced hybrid models (Dynex-QML with 8B and 70B base models) incorporating quantum optimization via Quadratic Unconstrained Binary Optimization (QUBO) formulations. The quantum-enhanced approach transforms the final attention layer of the LLM into a global constraint satisfaction problem solved via neuromorphic quantum annealing. Model performance was measured on a held-out test set of 500 notes using precision, recall, and -score metrics. Computational efficiency was quantified through end-to-end processing time. Results: The quantum-enhanced Dynex-QML-70B model achieved the highest overall performance with a macro-score of 0.855 (95% CI 0.823‐0.880), outperforming the stand-alone Llama-3.3-70B (0.726, 95% CI 0.704‐0.747), Dynex-QML-8B (0.733, 95% CI 0.709‐0.756), and Llama-3.1-8B (0.602, 95% CI 0.588‐0.615). Compared with Llama 3.3 70B, Dynex-QML (Llama 70B) improved macro-score by 0.128 (95% CI 0.091‐0.163; empirical 2-sided bootstrap

This week's Java roundup for September 7th, 2026, features news highlighting: new JEPs for ahead-of-time compilation and structured concurrency; GA releases of Jakarta CDI 5.0 and ADK for Kotlin 1.0; the September 2026 edition of Open Liberty; point releases of TornadoVM and RefactorFirst; a maintenance release of Micronaut; and first releases candidates of Groovy 6.0 and Gradle 9.8.

By Michael Redlich

Background: Accurate measurements of children’s digital technology use are essential for understanding its potential implications on health and well-being. Wearable cameras can provide such measurements, but image coding is a high burden for researchers. Machine learning–based object-recognition models have the potential to reduce this burden by identifying images containing technology. Objective: This study aims to evaluate the performance of an object recognition model, the Square Eyes model, as a screening tool for identifying technologies in wearable camera images among children for further human review, as well as to examine the potential influence of face-blurring methods on the model’s performance. Methods: This study used data collected on 48 children (aged 3‐14 y) during an approximately 1-hour laboratory session. The children performed various technology-related tasks while wearing a camera. A total of 221,226 images were coded by humans and processed through the Square Eyes model. The performance of the Square Eyes model as a screening tool was evaluated by (1) assessing agreement between the model and human coding; (2) evaluating the N-back algorithm, an algorithm embedded in the model aimed to flag images requiring human review; and (3) examining the potential influence of facial-blurring on model performance. Results: Humans detected technology in 92,745 (41.9%) images, and the Square Eyes model detected technologies with an overall accuracy of 78.0%. When considering specific technologies, agreement between the model and human coders was the highest for (n=19,148, 54.3%) and (n=8492, 44.5%) and lowest for smaller devices such as (n=2600, 31.3%) and (n=3685, 25.1%). The model’s N-back algorithm effectively flagged images that required further human review, with only 7144 (3.2%) images that were not flagged for screening containing a human-coded technology. An explorative analysis indicated that using a square face-blurring with border could have reduced the model’s ability to accurately detect technologies. Conclusions: The Square Eyes model demonstrated overall satisfying accuracy in detecting technologies and successfully flagged images that required further review by humans. These findings suggest that the model could be used as an effective screening tool for reducing the burden of human coding. However, the model could be improved to more accurately detect smaller devices, and the form of facial blurring in images should be considered.

Background: Chinese men who have sex with men living with HIV (MSMLWH) experience substantial psychological distress driven by minority stress and HIV-related challenges. However, culturally tailored digital mental health interventions that address HIV-specific maladaptive cognitive schemas and culturally specific psychosocial stressors remain scarce in China. Objective: This study aimed to systematically adapt an evidence-based cognitive behavioral therapy (CBT) intervention Effective Skills to Empower Effective Men (ESTEEM) into a WeChat (Tencent) Mini-Program–based intervention (iESTEEM) specifically for Chinese MSMLWH and to evaluate its preliminary feasibility and usability. Methods: We used a three-phase user-centered design approach guided by the Assessment, Decision, Adaptation, Production, Topical Experts, Integration, Training, and Testing (ADAPT-ITT) framework. The study proceeded in three phases: (1) a qualitative needs assessment using semistructured interviews with 20 MSMLWH (mean age 23.25, SD 3.08 years); (2) systematic intervention adaptation and platform development, including theater testing (n=5); and (3) a 2-week pilot study involving 10 MSMLWH and five counselors to evaluate feasibility, usability, and acceptability through focus groups and objective platform analytics. Results: Phase 1 identified 3 major themes of psychological distress: persistent health anxiety fueled by catastrophizing, intersectional stigma internalization, the disclosure dilemma, and intimacy barriers rooted in defectiveness and shame schemas. Participants also prioritized anonymity and bite-sized learning. Guided by these findings, iESTEEM was developed as a counselor-assisted, privacy-preserving WeChat Mini-Program incorporating HIV-specific scenarios, multimodal learning modules, and a back-end risk-alert system. During the 2-week pilot, participants logged into the platform 14.1 (SD 6.7) times per person and completed 134.3 (SD 103.1) minutes of learning activities; all participants accessed module 1, and 90% (9/10) accessed modules 2‐5. Anxiety scores decreased from 8.9 (SD 2.3) to 7.2 (SD 3.0), whereas depression scores remained stable. All participants expressed a willingness to continue using the program and to recommend it to peers. Participants and counselors endorsed its contextual relevance, privacy protections, and clinical utility. Conclusions: This study provides a theory- and evidence-informed model for culturally adapting digital mental health interventions for highly stigmatized populations. By integrating lesbian, gay, bisexual, transgender, and queer (LGBTQ)-affirmative CBT principles, HIV-specific adaptations, and a privacy-preserving, counselor-assisted WeChat Mini-Program, iESTEEM demonstrated promising preliminary feasibility, acceptability, and engagement among Chinese MSMLWH. These findings support the potential of culturally tailored digital interventions to expand access to psychological support for this stigmatized population in resource-constrained settings. Ongoing randomized controlled trials will further evaluate its efficacy, implementation outcomes, and mechanism of action. Trial Registration: Chinese Clinical Trial Registry ChiCTR2400080263; https://www.chictr.org.cn/showproj.html?proj=216926

Agoda migrated its 1.5 TB hotel Price Cache from 72 SQL Server shards to DragonflyDB to handle growing read and write volumes. The migration used staged dual reads, parity validation, gradual traffic shifting, and decentralized failover detection. Agoda reports an approximately eightfold reduction in P99 read latency, with two DragonflyDB clusters providing high availability.

By Leela Kumili

Alex Porcelli discusses the critical gap in enterprise AI: non-deterministic output and lack of accountability in high-stakes decisions. He shares how integrating DMN decision models with LLMs, agent skills, and NeMo guardrails creates auditable, deterministic agentic architectures - allowing business leaders to own decision logic while engineers maintain robust architectural governance.

By Alex Porcelli

Live translation has long been one of travel’s hardest unsolved problems: a single guide speaking to a mixed-language group, with no way to be understood by everyone at once. Vox Group, a 25-year-old guiding technology company operating in over 150 countries, has spent the past year rebuilding its AI-powered technology, Aura, to close that gap, now supporting simultaneous translation in up to 200 languages, live accessibility subtitles, and an AI companion designed to support guides rather than replace them. The update arrives as Vox celebrates its 25th Anniversary this month.

WHO IS VOX GROUP

Few outside the travel industry have heard of Vox Group, yet its technology underpins how millions of people experience a guided tour, powering cruise lines, tour operators, DMCs and destination experience partners in more than 150 countries. Founded 25 years ago on a simple radio link that let one guide be heard clearly by an entire group, the company has spent the years since expanding that idea, first into group guiding hardware, and more recently into Aura, its AI-powered guiding technology. Vox remains privately, family-led, an increasingly unusual profile among travel technology providers operating at this scale.

“Twenty-five years ago, we put a radio into a guide’s hand so that every guest could hear them,” says Elio Epifani, founder of Vox Group. “That principle has not changed. Aura begins with the guide and stays with the guide. What has changed is how much it can carry for them.”

THE TRANSLATION PROBLEM AI IS NOW SOLVING

Group tours have traditionally handled multiple languages in one of two ways: splitting visitors into separate language groups or adding a dedicated interpreter alongside the guide. Both add cost and complexity, and neither scale well for operators running mixed-nationality departures, increasingly the norm on cruise excursions and city tours. Aura’s translation library has grown from 50 to 200 languages over the past year, with live simultaneous translation now running in up to four languages at once on a single tour. Guests need no smartphone, app or download to take part, audio reaches them over the same radio infrastructure Vox has used for 25 years.

AN AI COMPANION BUILT TO SUPPORT GUIDES, NOT REPLACE THEM

Aura’s most significant update is its AI companion, a system designed to sit alongside the guide rather than in front of them. Rather than generating commentary from the open internet, it draws only on operator-approved content and verified sources, surfacing answers the moment a guest asks something unexpected, along with local context such as place names or regional sayings. It reflects a wider question the artificial intelligence sector is grappling with well beyond travel: where automation adds genuine value, and where it risks replacing the human expertise it was meant to support. Vox’s approach has been to keep the guide as the primary voice, using AI to extend what one person can reasonably be expected to know, rather than to generate the tour itself.

“One guide speaks once, and every guest receives a tour made for them in their own language, run from a single app,” says Fabio Primerano, Chief Executive Officer of Vox Group. “It gives guides more to work with and operators more to sell.”

ACCESSIBILITY BUILT IN, NOT BOLTED ON

Aura also addresses a gap much of the guiding technology on the market has overlooked. For guests who are deaf or hard of hearing, live subtitles are sent directly to their own phone as the guide speaks, letting them follow the tour in real time alongside the rest of the group rather than reading a summary afterwards. No separate equipment is issued, and no guest is visibly marked out as needing extra support, an accessibility approach built into the core product rather than offered as an add-on.

MULTIPLE CHANNELS, ONE DEPARTURE

Aura runs multiple commentary channels in parallel on a single tour, standard commentary, a historical deep dive, a children’s version, or a translated feed, so a family, a subject specialist and a first-time visitor can each follow a version suited to them without the guide repeating themselves. For operators, that means one departure can now carry visitors who would previously have needed splitting into separate language groups, with the tour adapting to each guest as it runs.

WHAT IT SIGNALS FOR AI IN TRAVEL

Aura’s evolution mirrors a pattern playing out across natural language processing applications more broadly: real-time translation and accessibility tools that once required specialist hardware or dedicated staff are increasingly built into existing infrastructure. For an industry still working out where AI adds value without eroding the human interactions travellers pay for, Vox’s approach, extending the guide rather than replacing them, offers one working answer.

Aura is available now for cruise lines, tour operators, DMCs and destination experience partners. voxtours.com/aura.

The post How Vox Group’s AI-Powered Technology Is Solving Real-Time Translation for Group Travel appeared first on AI News.

ESP32 Bit Pirate project integrates multi-protocol debugging within a browser environment using HTML5 APIs. It enables users to install firmware and interact with microcontrollers like the ESP32-S3 directly from the browser. The platform supports various digital and wireless protocols while providing hands-on guidance through practical recipes for tasks such as memory dumping and signal analysis.

By Olimpiu Pop

arXiv:2609.11977v1 Announce Type: new Abstract: Co-work agents execute complex workflows that combine information gathering, tool use, coding, and file manipulation across many model invocations. Because cost and latency accumulate over the full episode, their practical value depends not only on peak capability but also on how efficiently that capability is delivered. Yet many steps in everyday work emphasize state tracking, coordination, recovery, and follow-through rather than frontier-scale reasoning. We present Occamy-1.0, a cost-efficient co-work model obtained by further training the post-trained Qwen3.6-35B-A3B checkpoint. We construct execution-grounded data and environments, capture replayable long-horizon trajectories across multiple harnesses, and use staged post-training to develop and consolidate complementary execution capabilities. Across a broad suite of co-work benchmarks, Occamy-1.0 is consistently among the strongest comparably sized models and remains competitive with substantially larger frontier systems on several tasks. Under our stated evaluation and pricing protocol, its aggregate performance across four representative benchmarks places it at the low-cost knee of the observed cost--performance Pareto frontier. Supporting evaluations in tool calling, coding, and instruction following further show that this specialization preserves broad agentic capability. We release the model weights and a subset of the training data to support research on practical co-work agents and agentic post-training.

arXiv:2609.12035v1 Announce Type: new Abstract: Cardiovascular diagnosis rests on integrating complementary modalities, like ECG, echocardiography, chest radiographs, and clinical variables, each capturing distinct but correlated aspects of cardiac physiology. Yet most medical foundation models remain modality-specific, combining modalities only for finetuning or post-training. This discards the cross-modal evidence clinicians naturally integrate and ignores the structure within each modality. We introduce Latent-Attention Masked Autoencoders (LAMAE), a multimodal, structure-aware masked autoencoder that jointly learns patient-level representations during self-supervised pretraining. Rather than fusing modalities post hoc, LAMAE exchanges information directly in the latent space through a shared latent-attention module operating over a study-view-entity hierarchy, enabling aggregation of variable observations and graceful handling of missing modalities. Pretrained on over 1.2 million MIMIC-IV hospital stays, LAMAE outperforms modality-specific pretraining and strong contrastive and vision-language baselines across multimodal hospital-stay tasks, such as in-hospital mortality, ICD-10 and DRG coding, and length of stay, while remaining competitive on unimodal tasks. These gains persist even when only a single modality is available at test time, showing that modeling both intra- and inter-modal structure yields more robust, transferable representations.

arXiv:2609.12115v1 Announce Type: new Abstract: Phase-resolving wave models such as FUNWAVE-TVD are the accuracy standard for nearshore dynamics, resolving the shoaling, refraction, and breaking of individual waves, but their cost rules them out for the ensembles, uncertainty quantification, and real-time warning that operational forecasting demands. Neural operators promise solver-level accuracy at a fraction of that cost, yet on wave-dominated fields the accurate ones are large: hybrid spectral-convolutional operators such as U-FNO (the strongest baseline in our study after DU-NO) buy their fidelity with tens of millions of parameters. We introduce DU-NO (Double U-shaped Neural Operator), a multiscale U-shaped spectral operator that attaches lightweight convolutional U-Net branches only at its two shallowest encoder and decoder levels. The placement follows a sampling argument: high-wavenumber content exists only on fine grids, so the local, full-band pathways go where that content lives, while the coarse, band-limited levels stay purely spectral. A depth-decaying mode schedule holds the model to 3.64M parameters, an order of magnitude below U-FNO. On our publicly released FUNWAVE-TVD benchmark, DU-NO attains the best autoregressive rollout error of six identically trained architectures, improving on U-FNO by 14.9% with 10.8x fewer parameters, and a frequency-band analysis shows the gain holds across all bands, including the high-wavenumber band where truncated-spectral operators collapse. Parameter-matched controls confirm the gain is architectural: rescaled to the same 3.6M budget, the best baseline still trails DU-NO by 28.6%. The advantage carries beyond nearshore waves: DU-NO matches the strongest baselines on 2D Navier-Stokes and wins clearly on PDEBench shallow-water rollouts. Code, trained models, and evaluation artifacts are available at https://anonymous.4open.science/r/duno-code-5A7B/.

arXiv:2609.12116v1 Announce Type: new Abstract: Editing a knowledge graph embedding (KGE) model to promote a desired answer can displace correct answers from the returned list. Locality tests based only on facts that reuse the edited parameter can miss this ranking effect. We introduce a common rank-displacement audit at three scopes: facts supported by the edited parameter, other correct answers to the target query, and correct answers across queries with the same relation. We also derive dimensional and geometric conditions for an update to improve the target while exactly preserving selected scores. On FB15k-237 with DistMult and ComplEx, direct promotion always moves the target into the top ten, but does so without damage in only 23.0--23.2\% of edits. Strict preservation causes no measured damage, yet succeeds in only 1.3--1.4\%. Support-regularized entity editing gives the highest joint success, 36.3--37.7\%, while rank-truncated preservation reaches 32.8--34.7\% and reduces the mean number of displaced answers from about 14 to 1.2. Experiments across dimensions, scorers, ranking conventions, and a learned editor show that locality depends on both the protected scope and the editing mechanism. KGE editing should therefore report correction success together with the incidence and severity of rank displacement.

arXiv:2609.12265v1 Announce Type: new Abstract: Large Language Models (LLMs) are increasingly asked to reason over structured data such as graphs, yet how reliably they can carry out multi-step graph algorithms in language remains unclear. Existing evaluations tend to use simple tasks on small graphs, to score code generation rather than reasoning over the graph itself, or to fix a single input format. We introduce Graph Theory Bench (GT Bench), a benchmark covering 24 classical graph problems in 44 task-structure settings, with over 100,000 examples across four representations: natural language, structured language, adjacency list, and adjacency matrix. Evaluating eight LLMs on GT Bench shows that accuracy is strongly tied to the input representation, that the best representation shifts with graph density, size, and topology as well as with the model, and that this sensitivity persists, attenuated, in the strongest reasoning models. Building on these observations, we propose the Graph Theory Agent (GTA), which pairs a preference-trained representation selector with plan-and-decompose scaffolding around a frozen executor LLM. GTA lifts Phi-4 from 53.5% to 69.1% on the benchmark's easy split and from 33.0% to 41.5% on its hard split, outperforming eight prompting and agent baselines, and transfers without retraining to GraCoRe and NLGraph. Code for benchmark generation and evaluation: https://github.com/xzx34/GTA. The project homepage is available at https://xzx34.github.io/gta/.

arXiv:2609.12267v1 Announce Type: new Abstract: Learning user-defined concepts as constraint networks has been extensively studied in the constraint acquisition (CA) literature. However, existing approaches typically rely on intensive interactions with a human oracle, making the learning process costly in terms of time and number of queries. In this paper, we propose a neuro-symbolic framework for automatic CA that significantly reduces user involvement by introducing neural Oracle Transformer models which learn to emulate user responses and to generalize conceptual knowledge. Trained on previously available examples, the learned oracle interacts with a dedicated CA engine, FastCA, which systematically refines the oracle's responses into a sound, consistent, and interpretable constraint network. This neuro-symbolic interaction enables the recovery of structured symbolic models from data without prior domain knowledge. Our results demonstrate that this neuro-symbolic interplay effectively aligns data-driven pattern recognition with symbolic reasoning, offering a robust approach to automating model construction in combinatorial domains.

arXiv:2609.12320v1 Announce Type: new Abstract: Traditional large language models (LLMs) are scoped to individual user sessions, limiting their knowledge to a single conversation and preventing them from learning user preferences that evolve over time. Existing agentic memory systems address this limitation but generally operate at the individual-user level, restricting the public knowledge that could be shared across users to improve downstream responses. We introduce AIM (Agentic Interoperable Memory), a unified, privacy-aware memory framework that enables multi-agent, multi-user LLM systems to persistently manage private and shared memory. AIM dynamically classifies information as private, scoped to one user and inaccessible to others, or public, accessible to all users. It enforces index-level access controls so that private memories are retrievable only by their owner, protecting sensitive data while allowing beneficial shared knowledge to improve coordination and consistency. We also introduce MUMBench (Multi-User Memory Benchmark), a dataset of multi-user interactions containing private and shareable information across four domains. To our knowledge, MUMBench is the first public dataset designed to evaluate multiple memory operations, including retrieval, creation, update, and deletion, in a multi-user environment. Across three independent runs on MUMBench, AIM achieves 96.0% visibility classification accuracy, 58.8% strict operation accuracy, and 70.5% state-aware operation accuracy.

arXiv:2609.12373v1 Announce Type: new Abstract: Persona drift remains a central challenge for personalized language models, as user profiles evolve over long interactions rather than remain permanently fixed. Models must therefore revise persistent persona states when preferences genuinely change, while avoiding updates driven by transient, ambiguous, or unresolved observations. We propose CORE, which separates turn-local evidence from persistent persona-state revision and selectively updates grounded user preferences through uncertainty-aware belief revision. We also introduce PERSIST, a held-out post-anchor benchmark for persona-state robustness under sequential interaction stress, covering ambiguity, conflict, and controlled social influence. Across ALOE, PersonaChat, and PERSIST, CORE improves personalized alignment and robustness, with complementary gains in normalized closed-slot state fidelity. Human evaluation and mechanistic controls further support explicit update control beyond stronger generation or persistent memory alone.

arXiv:2609.12394v1 Announce Type: new Abstract: Mobile GUI agents are shifting from multi-module frameworks to native models trained end-to-end, yet industrial deployment faces three persistent gaps. Sandbox training produces a distribution mismatch with production environments; expensive real-device failures remain underutilized; and fixed benchmarks saturate, losing the power to guide iteration. We present BlueLM-GUI, a 35B-A3B mobile GUI agent built as a real-device-centric flywheel that closes these gaps through three principles. Every Sample Matters: a dual-track pipeline with Heterogeneous Triple-System Consensus evaluation and an Error Correction \& Derivation Module salvages every trajectory into usable supervision. Every Rollout Is Real: a three-stage recipe---continual pre-training, supervised fine-tuning, and agentic reinforcement learning on hundreds of real phones---grounds every rollout in real production environments, so the capability the model learns transfers directly to deployment. Every Query Evolves: a quota-driven benchmark methodology with three orthogonal axes enables precise attribution and allows the benchmark to be systematically upgraded as the model improves. BlueLM-GUI achieves 87.4 on MobileGUI-VBench, surpassing the best closed-source model by 5.1 points, and 84.9 on AndroidWorld, the best result among open-source models and competitive with closed-source models. These results demonstrate that grounding model training and iterative improvement in both real devices and the three Every principles yields strong, robust, and transferable mobile GUI capability.

arXiv:2609.12399v1 Announce Type: new Abstract: Generative recommendation (GR) relies on large-beam decoding to generate hundreds of candidate items, creating a new scaling challenge for recurrent linear attention. Existing linear attention serving systems either materialize a full recurrent state for every beam or repeatedly replay shared history, incurring substantial memory and traffic overhead. To address this, we present OneLA, a linear-attention decoding framework that exploits the shared prompt and short divergent suffixes of GR workloads. Specifically, OneLA represents all beam states using a single shared prompt-derived state and compact, append-only records of their divergent transitions. Using this representation, OneLA computes only the state information required at each decoding step, without reconstructing a full recurrent state for every beam. Furthermore, OneLA uses a lightweight ancestry index to track the transition records that make up each beam's history, allowing beams to be updated without moving or copying existing records. A fused GPU kernel further reuses the shared state across beams. Our analysis shows that OneLA achieves 1.54-2.46x end-to-end decode speedups while substantially reducing recurrent-state memory use and data movement.