Thursday, July 23, 2026
- CASIVIBOTv8CASIVIBOT is China's first industrial embodied quality inspection robot, developed by CASIVISION (Zhongke Huiyuan Vision Technology (Luoyang) Co., Ltd.)
- AEI Robot Alicev6Alice is a series of humanoid robots developed by AeiROBOT Co., Ltd., a South Korean robotics startup headquartered in Ansan, Gyeonggi-do.
- Ameca (robot)v7Ameca is a humanoid robot developed by Engineered Arts, a British robotics company based in Falmouth, Cornwall, United Kingdom, that the company markets as "the world's most advanced social humanoid robot."…
- Naver Labs Ambidexv9Naver Labs Ambidex (stylized as AMBIDEX) is a dual-arm, cable-driven robot system developed by Naver Labs, the research and development subsidiary of Naver Corporation, South Korea's largest internet company.
- AgiBot A2 Maxv6The AgiBot A2 Max (Chinese: 远征A2 Max, Yuanzheng A2 Max) is a heavy-duty bipedal humanoid robot developed by AgiBot (Zhiyuan Robotics) of Shanghai, China, and it is the largest and most powerful variant in the…
- Agile Onev8Agile One (stylized as Agile ONE) is an industrial humanoid robot developed by Agile Robots SE, a Munich-based robotics company originally spun off from the German Aerospace Center (DLR).
- Tangible Robots Eggiev7Eggie is a wheeled humanoid robot developed by Tangible Robots, a robotics startup based in Palo Alto, California.
- Svaya Robotics Bimanualv5The Svaya Robotics Bimanual is a semi-humanoid, dual-arm collaborative robot designed and manufactured by Svaya Robotics, an Indian robotics company headquartered in Hyderabad, Telangana.
- Adam SPv6Adam SP is a full-size humanoid robot developed by PNDbotics, a Chinese robotics company specializing in full-stack humanoid robot design and manufacturing.
- 1X EVEv61X EVE is a wheeled humanoid robot developed by 1X Technologies (formerly Halodi Robotics), a Norwegian-American robotics company founded in 2014.
- AgiBot A2 Ultrav7The AgiBot A2 Ultra (Chinese: 远征A2 Ultra, Yuanzheng A2 Ultra) is the perception-enhanced flagship variant of the AgiBot A2 interactive-service humanoid robot
- AgiBot A2v6The AgiBot A2 (Chinese: 远征A2, Yuanzheng A2) is a full-size bipedal humanoid robot developed by AgiBot, a Chinese robotics company headquartered in Shanghai.
- AIDOLv5AIDOL is a humanoid robot developed by Artificial Intelligence Dynamic Organism Lab (commonly abbreviated as AIDOL or Idol), a Russian robotics startup headquartered in Moscow.
- TeknTrash ALPHAv5ALPHA (short for Automated Litter Processing Humanoid Assistant) is a humanoid robot developed by TeknTrash Robotics, a United Kingdom-based company specializing in AI-powered robotics and motion intelligence…
- AgiBot X2v6The AgiBot X2, also known as the Lingxi X2 (灵犀X2), is a compact bipedal humanoid robot developed by AgiBot (Zhiyuan Robotics), a Shanghai-based Chinese robotics company, and unveiled on March 11, 2025.
- AgiBot A2-Wv7The AgiBot A2-W (Chinese: 远征A2-W, Yuanzheng A2-W) is a wheeled dual-arm mobile manipulator built for factory automation by AgiBot (Zhiyuan Robotics), a Shanghai-based Chinese robotics company.
- NEURA Robotics 4NE-1v7The NEURA Robotics 4NE-1 (pronounced "for anyone") is a cognitive humanoid robot built by the German company NEURA Robotics, headquartered in Metzingen, near Stuttgart, in Germany.
- Hexagon AEONv6Hexagon AEON is an industrial humanoid robot developed by the Robotics Division of Hexagon AB, a Swedish multinational technology company known for precision measurement, sensor technology, and industrial…
- PNDbotics Adam Litev6PNDbotics Adam Lite is a full-size bipedal humanoid robot developed by PNDbotics, a Chinese robotics company specializing in full-stack humanoid robot design and manufacturing.
- ROBOTIS AI Workerv7The ROBOTIS AI Worker, formally designated FFW (Freedom From Work), is a semi-humanoid robot platform for industrial physical AI, built by ROBOTIS Co., Ltd., a South Korean robotics company founded in 1999 and…
- 4NE-1v6*This article is a detailed summary. For an even fuller treatment, see also NEURA Robotics 4NE-1.*
- AgiBot A2 Litev6The AgiBot A2 Lite (Chinese: 远征A2 Lite) is the most affordable full-size bipedal humanoid robot sold by AgiBot (Zhiyuan Robotics), a Shanghai-based Chinese robotics company, priced at approximately $44,560 and…
- SHAP (SHapley Additive exPlanations)v6SHAP (SHapley Additive exPlanations) is a game-theoretic method that explains an individual machine learning prediction by assigning each input feature a numerical value representing how much it pushed that…
- Grokkingv5Grokking, also called delayed generalization, is a phenomenon in deep learning where a neural network first memorizes its training data (achieving near-perfect training accuracy but random-level test…
- LIMEv7LIME (Local Interpretable Model-Agnostic Explanations) is a technique for explaining individual predictions of any black-box machine learning classifier or regressor by approximating the model locally with a…
- Double Descentv4Double descent is a phenomenon in machine learning and statistical learning theory in which a model's test error, plotted against increasing model complexity, first traces the classical U-shaped bias-variance…
- Discount Factorv5The discount factor, almost always written as the Greek letter $$\gamma$$ (gamma), is a scalar hyperparameter in reinforcement learning that controls how much an agent values future rewards relative to…
- Singular value decompositionv8Singular value decomposition (SVD) is a matrix factorization that writes any real or complex m x n matrix $$A$$ as the product $$A = U \Sigma V^\top$$, where U and V are orthogonal matrices (the left and right…
- Dimensionality reductionv6Dimensionality reduction is the process of transforming data from a high-dimensional space into a lower-dimensional representation that retains as much of the meaningful structure of the original data as…
- t-SNEv8t-distributed stochastic neighbor embedding (t-SNE) is a nonlinear dimensionality reduction technique used primarily for visualizing high-dimensional data in two or three dimensions.
- Direct Preference Optimization (DPO)v8Direct Preference Optimization (DPO) is a method for aligning large language models with human preferences that replaces the multi-stage reinforcement learning from human feedback (RLHF) pipeline with a single…
- Curse of Dimensionalityv6The curse of dimensionality is the set of problems that arise when data has a large number of features (dimensions): as dimensions increase, the volume of the space grows exponentially, the available data…
- Expectation-Maximization (EM) Algorithmv8The Expectation-Maximization (EM) algorithm is an iterative method for finding maximum likelihood or maximum a posteriori (MAP) estimates of the parameters of statistical models that involve latent…
- Estimator (tf.estimator)v5tf.estimator is a high-level TensorFlow API that encapsulates the complete lifecycle of a machine learning model, including training, evaluation, prediction, and export for serving.
- UMAP (Uniform Manifold Approximation and Projection)v5UMAP (Uniform Manifold Approximation and Projection) is a nonlinear dimensionality reduction technique that compresses high-dimensional data into a low-dimensional map (typically 2 or 3 dimensions) while…
- Grad-CAMv5Grad-CAM (Gradient-weighted Class Activation Mapping) is a technique for producing visual explanations from convolutional neural network (CNN) models by using the gradients of a target class flowing into the…
- Rejection samplingv5Rejection sampling, also called the accept-reject method or the acceptance-rejection method, is a Monte Carlo technique that draws independent samples from a hard-to-sample target distribution p(x) by…
- AdvBenchv6AdvBench (Adversarial Behavior Benchmark) is a red-teaming benchmark dataset for measuring how easily an aligned large language model can be pushed into producing harmful or objectionable content
- InfiniteBenchv7InfiniteBench (stylized as ∞Bench) is a long-context benchmark that tests whether large language models (LLMs) can genuinely process and reason over inputs longer than 100,000 tokens, using 12 tasks that span…
- MACHIAVELLI (benchmark)v4MACHIAVELLI is a benchmark for evaluating the ethical behavior of AI agents in text-based interactive environments.
- EgoSchemav6EgoSchema is a diagnostic benchmark for evaluating very long-form video language understanding, introduced by Karttikeya Mangalam, Raiymbek Akshulakov, and Jitendra Malik at UC Berkeley.
- HarmBenchv5HarmBench is a standardized evaluation framework for automated red teaming and robust refusal of large language models (LLMs).
- RULER (benchmark)v6RULER is a synthetic benchmark from NVIDIA that measures the real, usable context window of large language models (LLMs) by testing them on 13 tasks across four categories (retrieval, multi-hop tracing…
- BBQ (Bias Benchmark for QA)v5BBQ (the Bias Benchmark for QA) is a hand-built evaluation dataset that measures whether a question answering (QA) language model relies on social stereotypes when it answers.
- JailbreakBenchv5JailbreakBench is an open-source robustness benchmark for evaluating jailbreak attacks and defenses against large language models (LLMs).
- LongBenchv8LongBench is a benchmark suite for evaluating the long-context understanding capabilities of large language models (LLMs).
- Depthwise Separable CNNv4A depthwise separable convolution is a factorized form of convolution that decomposes a standard convolutional operation into two sequential steps: a depthwise convolution and a pointwise convolution.
- Earth Mover's Distancev7Earth Mover's Distance (EMD), also known as the Wasserstein-1 distance, Kantorovich-Rubinstein metric, or Mallows's distance
- TPU Boardv5A TPU board (Tensor Processing Unit board) is a printed circuit board (PCB) that houses one or more Tensor Processing Unit chips along with associated memory, power delivery, and interconnect components.
- SUPERBv4SUPERB, which stands for Speech processing Universal PERformance Benchmark, is a comprehensive evaluation framework designed to measure how well self-supervised learning (SSL) models generalize across a…
- ToxiGenv4ToxiGen is a large-scale, machine-generated dataset designed for adversarial and implicit hate speech detection.
- PR AUCv5PR AUC (Precision-Recall Area Under the Curve), also referred to as AUPRC or AUC-PR, is a classification evaluation metric that quantifies the area beneath a precision-recall curve.
- Empirical Risk Minimizationv5Empirical risk minimization (ERM) is the foundational principle of statistical learning theory: because the true risk (the expected loss over the unknown data distribution) cannot be computed
- Needle in a Haystack (NIAH)v7Needle in a Haystack (NIAH) is a long-context evaluation that measures whether a large language model can retrieve a single fact (the "needle") inserted at a controlled position inside a long body of text (the…
- CRUXEvalv5CRUXEval (Code Reasoning, Understanding, and eXecution Evaluation) is a benchmark designed to measure how well large language models can reason about, understand, and mentally execute short Python programs.
- PIQAv7PIQA (Physical Interaction Question Answering) is a benchmark dataset of roughly 21,000 binary multiple-choice questions that evaluates the physical commonsense reasoning abilities of natural language…
- IFEvalv6IFEval (Instruction-Following Evaluation) is a benchmark of 541 prompts that measures how reliably large language models obey explicit, machine-checkable instructions such as "write in more than 400 words,"…
- WritingBenchv5WritingBench is a comprehensive benchmark for evaluating the generative writing capabilities of large language models (LLMs) across diverse real-world writing tasks.
- HaluEvalv4HaluEval (Hallucination Evaluation) is a large-scale benchmark for measuring how well large language models (LLMs) can recognize hallucinated content, that is, text that conflicts with a source or cannot be…
- LibriSpeechv5LibriSpeech is a freely available corpus of approximately 1,000 hours of 16 kHz read English speech that serves as the standard benchmark for training and evaluating automatic speech recognition (ASR) systems.
- LAMBADAv5LAMBADA (LAnguage Modeling Broadened to Account for Discourse Aspects) is a benchmark dataset designed to evaluate the ability of computational language models to understand broad discourse context.
- BIG-Bench Hardv7BIG-Bench Hard (BBH) is a suite of 23 challenging tasks drawn from the BIG-Bench benchmark, selected because they are "the [tasks] for which prior language model evaluations did not outperform the average…
- TruthfulQAv7TruthfulQA is a benchmark designed to measure whether large language models (LLMs) generate truthful answers to questions.
- MMMU-Prov4MMMU-Pro is a rigorous benchmark for evaluating multimodal AI systems on college-level, expert questions that genuinely require seeing an image, built as a harder and more robust version of the original MMMU…
- CLIP Scorev7CLIP Score (also written CLIPScore or CLIP-S) is a reference-free automatic evaluation metric that measures how well a text caption matches an image, computed as the rescaled cosine similarity of the image and…
- MathVistav7MathVista is a benchmark for evaluating the mathematical reasoning capabilities of foundation models in visual contexts.
- FLORES-200v5FLORES-200 is a multilingual evaluation benchmark for machine translation systems, covering 200 languages across a wide range of language families, scripts, and resource levels.
- BoolQv4BoolQ (Boolean Questions) is a natural language processing benchmark dataset of 15,942 naturally occurring yes/no question answering examples, each pairing a real Google search query with a Wikipedia passage…
- PubMedQAv6PubMedQA is a biomedical question answering dataset and benchmark that evaluates whether machine learning models can answer yes/no/maybe research questions using evidence from PubMed abstracts.
- GAIA benchmarkv6GAIA (General AI Assistants) is a benchmark for evaluating general-purpose AI agents and assistants on real-world tasks that require reasoning, web browsing, file handling, and multimodal understanding.
- LegalBenchv7LegalBench is a collaboratively constructed benchmark for measuring legal reasoning in large language models (LLMs)
- ZebraLogicv6ZebraLogic is a benchmark for evaluating the logical reasoning capabilities of large language models (LLMs).
- Viggle AIv4Viggle AI is an artificial intelligence-powered character animation and video generation platform developed by WarpEngine Canada Inc. The platform enables users to animate static images into dynamic videos…
- Photoroomv6Photoroom is an AI-powered photo editing platform headquartered in Paris, France, specializing in background removal, product photography, and generative image editing.
- TriviaQAv6TriviaQA is a large-scale reading comprehension and question answering dataset of over 650,000 question-answer-evidence triples, introduced in 2017 by Mandar Joshi, Eunsol Choi, Daniel S. Weld
- Basetenv8Baseten is an inference platform for deploying, serving, and scaling machine learning models in production.
- Berkeley Function Calling Leaderboardv5The Berkeley Function Calling Leaderboard (BFCL) is the standard benchmark for measuring how accurately large language models (LLMs) invoke functions, APIs, and tools, created by the Gorilla project at UC…
- CommonsenseQAv4CommonsenseQA is a multiple-choice question answering benchmark of 12,247 questions, introduced in 2019 by Alon Talmor, Jonathan Herzig, Nicholas Lourie, and Jonathan Berant
- Frechet Inception Distancev5The Frechet Inception Distance (FID) is the standard metric for measuring the quality of images produced by generative models: it computes the Frechet distance between two multivariate Gaussian distributions…
- MedQAv4MedQA is a large-scale, open-domain medical question answering benchmark of multiple-choice questions taken from real medical licensing examinations, introduced by Di Jin and colleagues at MIT in 2020.
- AlpacaEvalv6AlpacaEval is an automatic evaluation framework for instruction-following large language models (LLMs) developed by Stanford University's Tatsu Lab
- AgentBenchv6AgentBench is a multi-dimensional benchmark for evaluating large language models (LLMs) as autonomous agents across eight distinct interactive environments
- CodeContestsv4CodeContests is a competitive programming dataset created by Google DeepMind for training and evaluating machine learning models on algorithmic problem-solving tasks.
- Kaiber AIv7Kaiber AI is a creative technology company that builds AI video generation tools for musicians, visual artists, and content creators.
- Covariant (company)v8Covariant, Inc. (originally Embodied Intelligence) is an American artificial intelligence company that builds AI software for robotic manipulation and warehouse automation, founded in October 2017 by UC…
- Alignment Research Centerv8The Alignment Research Center (ARC) is a Berkeley, California nonprofit research organization, founded in April 2021 by Paul Christiano
- Qodov6Qodo (formerly CodiumAI or Codium) is an AI-powered code integrity platform that provides automated code review, test generation, and code quality tools for software developers.
- Anyscalev10Anyscale is an American technology company, founded in 2019 by the creators of Ray at the University of California, Berkeley, that develops the Anyscale Platform: a fully managed compute service built on Ray…
- Taskadev7Taskade is an AI-powered workspace platform for project management, task automation, and team collaboration.
- Mem AIv6Mem AI is an AI-powered note-taking and knowledge management platform developed by Mem Technologies, Inc. (commonly referred to as Mem or Mem Labs).
- tl;dvv3tl;dv (short for "too long; didn't view") is an artificial intelligence-powered meeting recording, transcription, and intelligence platform that works with Zoom, Google Meet, and Microsoft Teams.
- Cogramv6Cogram is an artificial intelligence platform designed for the architecture, engineering, and construction (AEC) industry.
- Center for AI Safetyv8The Center for AI Safety (CAIS) is an American nonprofit research and advocacy organization based in San Francisco, California, founded in 2022 to reduce societal-scale risks from artificial intelligence.
- Sourcegraph Codyv7Sourcegraph Cody is an enterprise AI coding assistant built by Sourcegraph that combines large language models with the company's code-search and code-graph technology to deliver context-aware chat…
- Wordtunev6Wordtune is an artificial intelligence-powered writing and reading assistant developed by AI21 Labs, an Israeli AI company.
- Pieces for Developersv7Pieces for Developers is an AI-powered developer productivity tool created by Mesh Intelligent Technologies
- Rytrv6Rytr is an artificial intelligence-powered writing assistant platform designed to help individuals and businesses generate short-form and medium-form written content.
- Sudowritev6Sudowrite is an artificial intelligence-powered writing assistant designed specifically for fiction writers, novelists, screenwriters, and other creative storytellers.
- Continue (software)v8Continue is an open-source AI code assistant that integrates directly into code editors, letting developers connect any large language model (LLM) and customize AI-powered coding features including…
- Exa AIv5Exa AI (formerly Metaphor) is an artificial intelligence company that builds a search engine designed specifically for AI applications.
