OrangeBot.AI Digest — 2026-10-22

Your deadline survival kit for office work

Proactive agent that understands what needs your attention

Mistral’s speech models for transcription, TTS and agents

A daily accountability call with other entrepreneurs.

Find bugs and vulnerabilities code review misses

AI Video Agent: Turn Anything into Videos via Chat

Hit record and hand off to AI, now in the new Loom app

Phone video to photoreal 3D & JSON-LD in an hour

Connect coding sessions to PRs, issues, quality, and ROI

Sync your calendars & connect them to ChatGPT, Claude & Co.

Omni-channel AI-first customer support (6 months free)

Live voice translation for chats, calls, meetings & YouTube

AI for your Apple Reminders

From messy filenames to shareable collections

Hire your AI Marketing Team

As streaming video generation advances, enabling real-time interaction with generated video is becoming increasingly important. We introduce StreamEval, a unified benchmark for interactive streaming video generation spanning five tasks: avatar dialogue, avatar performance, world generation, instruction editing, and reference editing. StreamEval comprises 300 scripted test cases lasting 15-300s, each specifying initial assets, incrementally arriving multimodal requests, and explicit requirements for state changes and content preservation. Performance is assessed along four evaluation dimensions: task execution, state consistency, visual fidelity, and audiovisual coordination, instantiated through ten task-specific metrics on a 0-100 scale. Evaluating 37 distinct models across 300 test cases, we find that high visual quality and identity retention do not necessarily coincide with reliable request execution or sustained state consistency. In particular, request execution remains challenging: the highest task-specific scores are 36.90 for Following among seven avatar-performance models, 25.93 for Following in text-conditioned world generation, and 15.00 for Edit Success among three reported reference-editing models. These findings identify reliable request execution under timing and preservation constraints as a central challenge for advancing omni-modal real-time interactive systems. The leaderboard is available at https://streameval.ai

Spatial intelligence requires agents to actively acquire visual evidence and integrate observations into a coherent representation of the three-dimensional world. Yet spatial question answering offers limited evidence of whether models can construct accurate metric maps through autonomous exploration. We introduce SpatialScout, a benchmark for evaluating vision-language model (VLM) agents on active object-level 3D mapping. SpatialScout comprises 80 diverse indoor scenes containing 3,140 target objects. Given an object inventory, RGB observations, camera calibration, and camera poses, agents control a camera to explore each scene and estimate object centers in a shared coordinate frame. The task requires coordinating viewpoint selection, cross-view association, geometric estimation, and map revision under bounded interaction budgets. Evaluation jointly measures localization accuracy and target coverage through object-scale overlap scores that penalize missing predictions. Experiments with frontier VLMs reveal substantial limitations: the strongest evaluated model, Claude Opus 5.5, achieves a mean overlap score of 41.71%, with 47.23% of targets exceeding an intersection-over-union (IoU) threshold of 0.5 and only 3.25% exceeding 0.9. These results reveal a substantial gap between recovering coarse object layouts and constructing complete, precise metric maps. SpatialScout provides a quantitative testbed for developing agents that build reliable 3D spatial representations through active perception. Code is available at https://github.com/JesseZZZZZ/SpatialScout.

Despite recent progress in dexterous reference tracking, existing models focus primarily on single-hand manipulation and a limited range of interactions, leaving the bimanual dexterous behaviors common in real life unaddressed. The scarcity of contact-rich bimanual manipulation data and the increased complexity of coordinating contacts make developing a generalist bimanual tracker particularly challenging. Therefore, we present 1) DONIC, the first generalist bimanual tracker jointly trained across operation categories and human-motion sources, together with 2) BiHOI, a multi-view dataset of 1,575 sequences spanning 229 objects and emphasizing sustained bimanual interaction, where reconstructed interactions are retargeted into robot references. DONIC combines proximal policy optimization with behavioral cloning in a feedback loop that recycles verified successful executions into training demonstrations. An object-centered per-hand reward maintains object-motion feedback while supervising each hand according to its reference contact role. Clip-Centered Advantages remove reference-dependent advantage offsets while preserving within-trajectory differences. Together, these designs enable one single policy to learn coordinated manipulation across heterogeneous references. Without finetuning, a single checkpoint achieves 86.79% success on unseen trajectories of training objects and 57.32% on held-out objects, surpassing the state-of-the-art bimanual tracking baselines by 77.36 and 45.73 percentage points.

Large Language Model (LLM) agents can handle increasingly complex tasks, navigating long trajectories of observations and actions. As these trajectories grow, key-value (KV) cache I/O and storage become the dominant cost. Trajectories can also exceed the LLM's maximum context length, requiring control over active cache growth. Prior work uses context engineering to reduce cache size by trimming or summarizing history. However, modifying the trajectory prefix invalidates subsequent cached states and triggers costly re-prefilling of retained context. Task accuracy can also suffer when too little context is retained to limit re-prefill costs. We propose cache engineering as a new scheme. It allows agents to directly control their active KV cache without modifying the trajectory or re-prefilling retained context. Specifically, agents can proactively use drop to exclude selected cache from future attention and repos to adjust retained keys' position encodings. We formalize these cache operations and identify history dependence as a serving challenge: identical token sequences can yield different cache states under different cache operation sequences, making prefix matching insufficient for correct cache reuse. Thus, we co-design the AKV serving engine. It provides a stateless interface compatible with existing serving APIs, while internally managing stateful cache reuse, eviction, and restoration. Experiments show that a tool-response rolling-drop cache policy improves average benchmark scores by 1.7 points and reduces average cost by 27.7%, compared with conventional threshold-triggered summarization. With AKV, it achieves up to 1.2x the output throughput and a 1.5x speedup in P95 request latency. Recursive self-improvement of cache policies further reduces cost by 7.1% over expert design. Our code is at https://github.com/thu-nics/AKV.

Autoregressive video diffusion models generate videos chunk by chunk, yet producing plausible continuations does not guarantee that they capture how scenes evolve. Under teacher forcing, strong temporal redundancy may allow local visual continuity to serve as a shortcut for current-chunk denoising, reducing the pressure to model visual evolution. We introduce Future Chunk Prediction (FCP), an auxiliary objective jointly optimized with teacher-forced denoising that predicts temporally separated chunks without access to the intervening visual context. Drawing supervision directly from training videos, FCP requires no additional pretrained visual encoder and leaves autoregressive inference unchanged. Our method achieves superior physical plausibility over standard teacher-forced autoregressive training with a lower overall training compute budget, highlighting the effectiveness and efficiency of future-chunk supervision for modeling visual evolution. Intriguingly, our visualizations suggest that FCP fosters a more structured representation space, which may be valuable for building autoregressive world models.

A login screen that appears after a tap on Sign in is expected; the same screen after a tap on View order is an attack. For GUI agents, safety is therefore a property of the transition rather than of the screen, and a monitor that inspects only screens can be defeated by reusing a legitimate one. Judging a transition requires an expectation of what should have followed the action. Existing GUI world models provide one, but they output it as text, code, or images, so checking it against the observed screen requires a second model to judge the two. We argue that a world model meant for verification should instead predict in the space in which observations are encoded, and present LGWM, a decoder-free, action-conditioned world model that predicts the representation of the next screen directly, trained without semantic annotation on 1.85M real GUI transitions. Verification reduces to a vector comparison, and the same signal reveals whether a mismatch is harmful. We evaluate on RSWT-BENCH, a diagnostic where each credential screen appears under both a legitimate and a hijacked transition, so detectors that see only the screen are at chance by construction. The training-free score reaches 0.987 AUC at 17 ms per decision, on par with the strongest closed-source VLMs and about ten AUC points above generative GUI world models at over three orders of magnitude lower latency. The residual direction reaches 0.953 AUC at separating harmful from benign violations, where prompted VLMs are near chance. Further analyses show that the prediction is a usable future state rather than an anomaly score. World models have mostly served as simulators or planners; our results point to a third role, verification, for which predicting in representation space is the natural design.

Agent learning is shifting from learning from human data toward learning through interaction with the world. However, scaling such learning directly in real-world computer systems is challenging and can introduce substantial security risks. We therefore propose Agent World Model in the Wild (WildAWM), a scalable pipeline for synthesizing controllable environments and tasks for training general computer-use agents (CUAs). Whereas existing approaches typically specialize in a single agent interaction interface, we construct executable environments that jointly support graphical user interfaces, model context protocol services, and command-line interfaces, allowing agents to flexibly combine visual interaction, tool use, and code execution. WildAWM is entirely powered by open-source models, demonstrating that high-quality computer-use environments can be generated without proprietary frontier models. Beyond environments, we address three recurring challenges in task synthesis: ensuring solvability, maintaining sufficient difficulty, and producing reliable verifiers, by introducing execution-based quality gates that combine heterogeneous agent rollouts, cross-judge consistency checks, and counterfactual probes of judgment robustness. The resulting pipeline synthesizes 341 long-horizon environment-task pairs that can require over 100 agent steps to solve and remain challenging even for frontier models, with GPT-6-Astra solving only 60% of them. We train two Qwen models of different sizes exclusively in these synthetic worlds. Both models achieve substantial gains across 8 out-of-distribution benchmarks spanning web, claw, and computer-use agents, demonstrating that learning in WildAWM's synthesized worlds generalizes broadly to unseen computer-use tasks. The code is publicly available at https://github.com/MSR-Orchard/WildAWM.

Multimodal code intelligence connects the symbolic world of programs with the visual world their execution creates. Interest in this capability is burgeoning, as models are increasingly asked to produce charts, webpages, animations, and semi-structured documents. Recently, proprietary models have pushed far ahead, producing interactive artifacts even from a text description or a screenshot alone. Yet their data and training remain closed, and open progress is held back by the quantity and quality of available data and by the pipelines and methods for producing it. Quality is the harder of the two, since a sample must be correct as a program, well built, and above all visually right once rendered. No existing signal covers all three. Execution often returns just a pass or fail, code similarity does not map to the visual output, and task-specific verifiers do not generalize. To push open progress further, we start from the data side with PrismaForge, a hierarchical data pipeline centered on a synthesized rubric that decomposes these factors into checkable criteria. The same rubric serves as both filter and reward: it curates the supervised corpus, then drives reinforcement learning on visual-coding tasks. Leveraging PrismaForge for data synthesis and filtering, we build PrismaCoder-2M, the most diverse open dataset for multimodal code intelligence and the largest to date. On the model side, we train PrismaCoder-9B and PrismaCoder-35B on this data and release both. They lead the open state of the art by a wide margin and rival leading proprietary systems. Our analyses further show that the proposed rubric-based pipeline improves multimodal coding performance across model backbones. We release our code and data at https://qiushisun.github.io/PrismaCoder-Home/.

Scaling data, diversity and model capacity has improved embodied learning in physical environments, where diverse experiences remain grounded in shared physical laws. Whether these gains extend to virtual worlds with different visual styles, dynamics, controls, and objectives remains unclear. We investigate this question through a controlled study of continued pre-training (CPT) across three axes: data volume, diversity, and model size. Our 120 training configurations span four model sizes from 0.8B to 27B, six data budgets from 50 to 1,600 hours, and five mixtures ranging from Minecraft-only data to diverse multi-world corpora. We assess performance and generalization using three-level validation losses and gameplay benchmarks from in-corpus worlds to unseen worlds. Our results show: (1) more data improves in-corpus performance, diversity improves transfer to related worlds with limited gains in distant worlds, and model size improves data efficiency but generalizes worse to distant worlds under single-world training; (2) we observe an early interface phase with broad cross-world benefits and a later world phase with gains narrowed to in-corpus and related worlds, and fit an empirical scaling law relating validation loss to data volume, diversity and model size; and (3) our models are competitive with Minecraft specialists and frontier models in 3D worlds, and perform best in real-time settings. However, the limits of imitation and persistent gaps in distant worlds suggest that building generalist embodied agents requires richer training stages beyond continued pre-training.

Post-hoc image explanations usually show where a model looks using spatial heatmaps or indicate which channels matter using global channel scores. These separate views are limiting for multispectral data because channel relevance is often spatially localized and physically meaningful. We introduce Spectral-Spatial Interpretation (SSI), a full-grid attribution method over position-channel players (g,c). For each spatial position, SSI fixes the observed surrounding context, defines a conditional game over the channels at that position, and applies the Shapley value to this local game. The result is a full spectral-spatial attribution matrix rather than a single heatmap or a global channel ranking. SSI has a clear local interpretation. It satisfies Shapley fairness within each position, and aggregating its channel attributions exactly recovers the position-level contextual contrast obtained by replacing that position with the baseline. SSI also differs from context-averaged Owen-style hierarchical accounting, which uses the same position-channel grouping but averages over external spatial coalitions. On HLS Burn Scar, SSI outperforms an MC-Owen estimator across retained attribution metrics while using only about 6% of its evaluation budget. Across CUB-200, EuroSAT, and HLS Burn Scar under both per-channel mean and zero baselines, SSI achieves the best joint deletion and insertion AUC over dense, flat, and rank-1 baselines, yields the strongest spectral hit rates and channel deletion, and leads spatial localization on both annotated datasets. Code is available at https://github.com/AtheavenMa/Spectral-Spatial-Interpretation.

Agent Skills improve LLM agents by making task knowledge reusable across sessions. Learning transferable Skills requires evidence about a procedure and its conditions of use, which a successful task attempt can leave untested. We introduce Socratic Skill Learning (SSL), which interleaves expert teaching with Student practice. An Educator guides investigation while the Student remains the sole environment operator and Skill author, able to question, test, and adapt advice. To test reuse beyond the learning task, we construct SkillTransferBench, with 35 task families and 408 executable cases, pairing public variation specifications with held-out evaluation in fresh solver sessions. On the main 22-family, 155-case cohort, Fixed-Window SSL reaches 0.801 utility, versus 0.723 without a Skill, 0.761 for adapted Trace2Skill (trajectory distillation), and 0.751 for adapted SKILL-KD (contrastive distillation). Across three model pairs on shared families, interleaved guidance outperforms expert authoring and post-hoc teaching. Process analysis links guidance to Student writing and package revision; a case study shows the Student testing and refining a proposed procedure for later reuse. As demand for new lessons falls, the cost of fixed-window review grows. Issue-Guided SSL tracks unresolved questions across Skill revisions to select and focus review, reaching 0.805 utility with 61.7% fewer Educator tokens and 16.9% fewer total acquisition tokens than Fixed-Window SSL on the main cohort.

Retrieval Augmented Generation (RAG) grounds large language models in external knowledge. Modern pipelines combine embedding, chunking, retrieval, reranking, query augmentation, context compression, and agentic verification, yet these techniques are typically evaluated one at a time. It remains unclear whether their isolated gains survive composition, whether retrieval quality and answer faithfulness favor the same design choices, and whether conclusions from small corpora hold at large scale. We pre-declare three hypotheses and introduce LASAGNA, an open-source framework for controlled RAG evaluation. We test 450 configurations across eight benchmarks and corpora ranging from 5K to 200K documents. Three findings emerge. First, individually strong techniques from different phases of a RAG pipeline do not necessarily compose: three retrieval techniques predict a +47% combined improvement but deliver only +17% when used together. Second, component gains stack differently for retrieval and faithfulness, and optimizing one can hurt the other. Third, the best composition changes as the corpus grows: the combination that wins on a small corpus may not remain best at large scale. These findings show that gains measured in isolation are not reliable predictors of realistic performance, and that new RAG techniques should be tested with the full surrounding stack. We call on researchers to do so and supply LASAGNA to simplify these experiments.

Quantitative research is an iterative loop: a researcher proposes a market hypothesis, implements it as an executable signal, tests it on historical data, and decides what to keep. Existing benchmarks score formulas, code, or individual trading decisions, leaving open whether large language models (LLMs) can complete this loop and produce strategies that work beyond the research period. We introduce Quant-Arena, a benchmark and public leaderboard for end-to-end quantitative research on real A-share market data. A model constructs features and stock-scoring code, receives feedback only on 2022--2023, and freezes one strategy for assessment on the sealed year 2024. A cash-account evaluator applies trading constraints and costs, including market impact estimated from a simulated order book. Scores are dominated by realized excess return and complemented by signal quality and the diversity of features used by the strategy. We evaluate nine frontier models, among which GPT-6 Astra scores highest; yet no model beats the equal-weight market in 2024. Quant-Arena is open to submissions from any model or agent harness.

Agentic robotics systems combine planners with robot harnesses to perform complex manipulation tasks. Recent work improves these systems through better harnesses and external memory, but relies on powerful closed-source models as planners and does not improve the planners' own capabilities. In contrast, coding agents have improved substantially by post-training the model itself. We therefore investigate whether post-training can also improve robot planners. However, applying it across robot harnesses requires accommodating different execution mechanisms, including VLA skill invocation and control-code generation. To this end, we introduce AgenticRoboRL, a framework for post-training robot planners across different harnesses. The training recipe combines supervised fine-tuning with multi-turn reinforcement learning. To accommodate different execution mechanisms, harness-specific adapters connect recorded planner interactions with task outcomes, enabling a common RL training pipeline while keeping execution components fixed. On LIBERO-PRO, AgenticRoboRL raises the success rate of a 9B-parameter planner from 18.1% to 71.0% without external memory, approaching the 72.1% achieved by Codex (GPT-5.5) with external memory. These results demonstrate post-training as an effective approach for improving robot planners. We will release the framework and trained planner checkpoints.

Autonomous agents powered by foundation models can reason, plan, and use tools, but their fixed weights and context limits hinder sustained learning from experience over long-term deployment. While external memory systems attempt to bridge this gap, they mostly act as passive retrieval caches, evaluated primarily by whether they can recall stored facts or past interactions. Whether memory can drive genuine continual learning remains an open question. In this paper, we introduce Memory-Centric Continual Learning (MCCL), arguing that memory should be judged by whether it improves future actions, not merely what it retrieves. MCCL has three main parts. First, a three-axis taxonomy organizes memory by where state is stored (substrates), what form it takes (assets), and how it is managed (policies). This defines a two-loop lifecycle that updates memory state in an inner loop while refining management policies in an outer loop based on task feedback. Second, an evaluation protocol measures retention under interference, adaptation to change, and task transfer to isolate genuine behavioral gains from simple context expansion. Third, a memory scaling framework models how agent performance scales with experience, storage capacity, and compute budgets across non-parametric and parametric memory. Together, these components reframe agent memory from passive lookup to continuous learning, offering a principled route toward self-improving autonomous agents.

Anthropic : Anthropic donates the Agent Skills open standard to the Agentic AI Foundation and says 70+ platforms have integrated it, after donating MCP to the AAIF in 2025 —  Today, we're donating Agent Skills, the open standard that gives AI agents specialized knowledge and workflows, to the Agentic AI Foundation (AAIF).

Kate Rooney / CNBC : Sources: OpenAI nears an up to $30B round led by investors from the UAE and Coatue at a $1.25T pre-money valuation, as an “opportunistic” bridge ahead of an IPO —  OpenAI is closing in on a funding round that could value the artificial intelligence giant at more than $1 trillion ahead …

Ivan Mehta / TechCrunch : Canva launches Craft, its first image generation model, trained on human-designed templates, available to all users, with plans to support third-party chatbots —  Design platform Canva released its first image generation model, called Canva Craft, developed by its research team to compete …

Camila Grigera Naón / Fortune : Microsoft and Alphabet join 14 other companies objecting to a plan by OKXICE, a venture of OKX and ICE, to trade blockchain-based versions of their shares —  A growing number of companies are pushing back against tokenized trading of their shares.  Microsoft and Alphabet are the latest …

Alfred Ng / Politico : Amazon issues refunds from a 2025 settlement, crediting “President Trump's Federal Trade Commission,” the first time an FTC refund has been tied to a president —  Millions of Americans are expected to receive a settlement check from Amazon credited to “President Trump's Federal Trade Commission” …

Robert Hart / The Verge : Microsoft says its GPT-5-powered Diagnostic Orchestrator uses agents that mimic panels of experts to diagnose conditions up to 4x more accurately than doctors —  'We're going to build something that is indistinguishable from human performance ... in the next couple of years,' Microsoft AI CEO Mustafa Suleyman tells The Verge.

The Information : Sources: Datadog offered to acquire Israeli cloud cybersecurity startup Upwind Security for nearly $4B; Upwind is finalizing a $300M round at a ~$3.8B valuation —  Enterprise software firm Datadog recently offered to buy Israeli cloud cybersecurity startup Upwind Security for nearly $4 billion …

Maria Diaz / ZDNET : Amazon rebrands its Kindle Unlimited subscription as Kindle+ and expands access to audiobooks, including titles narrated by AI —  Unlike narrated audiobooks from Audible and other providers, the audiobooks created with Virtual Voice in Kindle+ will instead be narrated by AI.  —  ZDNET's key takeaways

The Information : Sources: SoftBank's SB Energy is considering going public as soon as December or waiting until 2027, as it closely watches Anthropic's IPO, expected in November —  SB Energy, the SoftBank Group-controlled developer of a massive planned Ohio data center project for OpenAI …

Wall Street Journal : US officials say Trump asked Elon Musk to lift Starshield restrictions; Ukraine officials said it was not fully fixed Thursday but was working at higher speeds —  Military-grade system appeared to be working more normally on Thursday, soon after the call  —  WASHINGTON—President Trump …

Natalie Breymeyer / Axios : Chicago-based Formic, which provides robots-as-a-service automation to manufacturers, raised $50M, including a $40M Series B and $10M in asset-backed financing —  Formic, a physical AI commercialization platform, raised $50 million in financing, CEO Saman Farid tells Axios exclusively.

Jay Peters / The Verge : The Browser Company says it will shut down its Arc browser on December 22, launched in 2023, shifting attention to its AI-focused Dia browser, launched in 2025 —  New feature development for the innovative web browser ended last year.  Now The Browser Company is telling users to try Dia, Arc's replacement.

Todd Bishop / GeekWire : Microsoft says it is closing The Garage, which for 17 years gave employees worldwide space and tools for side projects and ran the company's Global Hackathon —  Microsoft is winding down The Garage as a companywide program, ending a 17-year run for the effort that gave employees space …

Lauren Feiner / The Verge : Filings: Nvidia, Anthropic, and OpenAI had their highest-spending lobbying quarter between July and September, spending $2.2M, $2.1M, and $1.5M, respectively —  Collectively, they spent nearly $6 million around cybersecurity, child safety, and more. … It was a busy summer for AI lobbyists.

Rocket Drew / The Information : Atlanta-based Tractian, which uses AI and sensors to predict machine failures and recommend maintenance plans, raised a $130M Series D at a $1.4B valuation —  Tractian, a 7-year-old physical AI startup based in Atlanta, Georgia, has raised $130 million at a valuation of $1.4 billion …