Amazon.com, Inc. ( AM-ə-zon) is an American multinational technology company based in Seattle, Washington, which focuses on e-commerce, cloud computing, digital streaming, and artificial intelligence. It is considered one of the Big Four companies in the U.S. information technology industry, along with Google, Apple, and Facebook. The company has been referred to as "one of the most influential economic and cultural forces in the world", as well as the world's most valuable brand.Amazon was founded by Jeff Bezos in Bellevue, Washington, on July 5, 1994. The company started as an online marketplace for books but expanded to sell electronics, software, video games, apparel, furniture, food, toys, and jewelry. In 2015, Amazon surpassed Walmart as the most valuable retailer in the United States by market capitalization. In 2017, Amazon acquired Whole Foods Market for US$13.4 billion, which substantially increased its footprint as a physical retailer. In 2018, Bezos announced that its two-day delivery service, Amazon Prime, had surpassed 100 million subscribers worldwide.Amazon is known for its disruption of well-established industries through technological innovation and mass scale. It is the world's largest online marketplace, AI assistant provider, live-streaming platform and cloud computing platform as measured by revenue and market capitalization. Amazon is the largest Internet company by revenue in the world. It is the second largest private employer in the United States and one of the world's most valuable companies. Amazon distributes downloads and streaming of video, music, and audiobooks through its Prime Video, Amazon Music, Twitch, and Audible subsidiaries. Amazon also has a publishing arm, Amazon Publishing, a film and television studio, Amazon Studios, and a cloud computing subsidiary, Amazon Web Services. It produces consumer electronics including Kindle e-readers, Fire tablets, Fire TV, and Echo devices. Its acquisitions over the years include Ring, Twitch, Whole Foods Market, and IMDb. The company has been criticized for various practices including technological surveillance overreach, a hyper-competitive and demanding work culture, tax avoidance, and for being anti-competitive.
AI agents may soon become capable of autonomously completing valuable, long-horizon tasks in diverse domains. Current benchmarks either do not measure real-world tasks, or are not sufficiently difficult to meaningfully measure frontier models. To this end, we present Terminal-Bench 2.0: a carefully curated hard benchmark composed of 89 tasks in computer terminal environments inspired by problems from real workflows. Each task features a unique environment, human-written solution, and comprehensive tests for verification. We show that frontier models and agents score less than 65% on the benchmark and conduct an error analysis to identify areas for model and agent improvement. We publish the dataset and evaluation harness to assist developers and researchers in future work at tbench.ai.
Agent Skills are structured packages of procedural knowledge that augment large language model (LLM) agents at inference time. Despite rapid adoption, there is no standard way to measure whether they actually help. We present SkillsBench, a benchmark whose current inventory contains 87 tasks across 8 domains paired with curated Skills and deterministic verifiers. Our latest aggregate evaluation runs the 87-task benchmark under matched no-Skills and curated-Skills conditions for 18 model-harness configurations. Curated Skills raise the average pass rate from 33.9
Reputation systems are central to online marketplace design, but they can fail in two related ways. First, buyers may generalize from individual transactions to the platform as a whole, creating reputational externalities across sellers. Second, public reputation measures may be biased or too coarse to let buyers accurately infer seller quality. We study these problems using detailed transaction-level data from eBay. We construct a measure of seller quality based on the share of transactions that generate positive feedback, including transactions for which buyers leave no feedback. Our measure predicts poor transaction outcomes and subsequent purchasing, even after controlling for reputation measures visible to buyers. We then evaluate a field experiment in which eBay incorporated this measure into its search-ranking algorithm. The treatment shifted buyers toward higher-quality sellers and increased subsequent purchasing without reducing contemporaneous conversion. The results highlight the need for richer models of platform markets that incorporate reputational externalities across sellers.
Vision-Language-Action (VLA) models aim to provide a single generalist controller for robots, but today's systems fall short on the criteria that matter for real-world deployment. Frontier models are closed, open-weight alternatives are tied to expensive hardware, reasoning-augmented policies pay prohibitive latency for their grounding, and fine-tuned success rates remain below the threshold for dependable use. We present MolmoAct2, a fully open action reasoning model built for practical deployment, advancing its predecessor along five axes. We introduce MolmoER, a VLM backbone specialized for spatial and embodied reasoning, trained on a 3.3M-sample corpus with a specialize-then-rehearse recipe. We release three new datasets spanning low-to-medium cost platforms, including MolmoAct2-BimanualYAM, 720 hours of teleoperated bimanual trajectories that constitute the largest open bimanual dataset to date, together with quality-filtered Franka (DROID) and SO100/101 subsets. We provide OpenFAST, an open-weight, open-data action tokenizer trained on millions of trajectories across five embodiments. We redesign the architecture to graft a flow-matching continuous-action expert onto a discrete-token VLM via per-layer KV-cache conditioning. Finally, we propose MolmoThink, an adaptive-depth reasoning variant that re-predicts depth tokens only for scene regions that change between timesteps, retaining geometric grounding at a fraction of prior latency. In the most extensive empirical study of any open VLA to date, spanning 7 simulation and real-world benchmarks, MolmoAct2 outperforms strong baselines including Pi-05, while MolmoER surpasses GPT-5 and Gemini Robotics ER-1.5 across 13 embodied-reasoning benchmarks. We release model weights, training code, and complete training data. Project page: https://allenai.org/blog/molmoact2
Benchmark quality is critical for meaningful evaluation and sustained progress in time series forecasting, particularly with the rise of time series foundation models (TSFMs). Existing benchmarks often have limited domain coverage or overlook real-world settings such as tasks with covariates. Their aggregation procedures frequently lack statistical rigor, making it unclear whether performance differences reflect true improvements or random variation. Many benchmarks lack consistent evaluation infrastructure or are too rigid for integration into existing pipelines. To address these gaps, we propose fev-bench, a benchmark of 100 forecasting tasks across seven domains, including 46 with covariates. Supporting the benchmark, we introduce fev, a lightweight Python library for forecasting evaluation emphasizing reproducibility and integration with existing workflows. Using fev, fev-bench employs principled aggregation with bootstrapped confidence intervals to report performance along two dimensions: win rates and skill scores. We evaluate various models on fev-bench and find that existing TSFMs often miss accuracy by ignoring covariates, pointing to a clear direction for future work.
