Thursday, July 23, 2026
- GPT Image 1v6GPT Image 1 (API identifier gpt-image-1) is a natively multimodal image generation model developed by OpenAI, integrated into ChatGPT on March 25, 2025, and released as a standalone API on April 23, 2025.
- Midjourney V7v3Midjourney V7 is the seventh major text-to-image generation model developed by Midjourney Inc. It launched in alpha on April 3, 2025, and became the platform's default model on June 17, 2025.
- Codestralv4Codestral is a family of code-specialized large language models developed by Mistral AI, beginning with Codestral 22B, released on May 29, 2024
- Mistral Largev7Mistral Large is the family of flagship large language models developed by Mistral AI, the Paris-based AI laboratory founded in 2023, and is the company's most capable general-purpose model line.
- Gemma 2v5Gemma 2 is a family of open-weights large language models developed by Google DeepMind and released starting June 27, 2024, in three parameter sizes: 2 billion (2B), 9 billion (9B), and 27 billion (27B).
- Wan 2.1v3Wan 2.1 (also written Wan2.1, from the Chinese Tongyi Wanxiang or 通义万象) is a family of open-weights text-to-video and image-to-video generation models that Alibaba's Tongyi Wanxiang team released and…
- Llama 3.3v6Llama 3.3 is an instruction-tuned, text-only large language model with 70 billion parameters that Meta released on December 6, 2024
- Recraft V3v3Recraft V3 is a text-to-image generation model developed by Recraft AI and released on October 30, 2024, that became the first model to reach the number-one position on the Artificial Analysis Text-to-Image…
- Phi-4v6Phi-4 is a 14-billion-parameter small language model developed by Microsoft Research and released in December 2024, designed to match or beat models several times its size on reasoning tasks by training…
- GRPOv5Group Relative Policy Optimization (GRPO) is a reinforcement learning algorithm for fine-tuning large language models that eliminates the separate critic (value) network used by PPO
- Grok 4v8Grok 4 is a large language model developed by xAI and released on July 9, 2025. It is the fourth major generation of the Grok model family and was positioned as xAI's most capable model to date at its release.
- Gemini 3v7Gemini 3 is the third major generation of the Gemini family of multimodal models from Google DeepMind, launched on November 18, 2025 with Gemini 3 Pro as the flagship and described by Google as "our most…
- Claude Haiku 4.5v6Claude Haiku 4.5 is a small, fast large language model developed by Anthropic and released on October 15, 2025, as the lightweight, high-speed member of the Claude 4.5 model family.
- ORPOv5ORPO (Odds Ratio Preference Optimization) is a preference alignment algorithm for large language models that merges supervised fine-tuning and preference alignment into a single training stage, eliminating the…
- KTOv3KTO (Kahneman-Tversky Optimization) is a method for aligning large language models with human feedback using only a binary signal of whether a model output is desirable or undesirable, rather than the paired…
- Grok 3v6Grok 3 is a family of large language models released by xAI, Elon Musk's artificial intelligence company, on February 17, 2025.
- Llama 3.2v7Llama 3.2 is a family of four open-weight large language models released by Meta on September 25, 2024, comprising lightweight 1 billion and 3 billion parameter text-only models for on-device AI and the 11…
- AWQ (Activation-aware Weight Quantization)v5Activation-aware Weight Quantization (AWQ) is a post-training quantization method for large language models that compresses weights to 4-bit (and optionally 3-bit) integers while keeping near-FP16 task…
- RadixAttentionv4RadixAttention is a KV cache management technique introduced in SGLang that uses a radix tree data structure to automatically share and reuse cached key-value tensors across inference requests.
- Mamba 2v4Mamba 2 is a state space model architecture introduced in the paper "Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Duality" by Tri Dao and Albert Gu
- RLVRv4Reinforcement Learning with Verifiable Rewards (RLVR) is a post-training paradigm for large language models in which the reward signal comes from a deterministic
- o4-miniv5OpenAI o4-mini is a compact reasoning model developed by OpenAI, released on April 16, 2025.
- PagedAttentionv5PagedAttention is a KV-cache memory management algorithm for serving large language models that applies the virtual-memory paging technique used by operating systems to the GPU, eliminating memory…
- Flash Attention 3v4Flash Attention 3 (FA3, styled FlashAttention-3) is the third generation of the FlashAttention algorithm
- GPTQv4GPTQ (Generative Pre-trained Transformer Quantization) is a one-shot post-training quantization method that compresses the weights of large language models to 3 or 4 bits using approximate second-order…
- Veo 3v8Veo 3 is a video generation model developed by Google DeepMind and announced at Google I/O on May 20, 2025, and it is the first commercially available video generation model to natively produce synchronized…
- Qwen3v7Qwen3 is the third-generation family of large language models developed by the Qwen Team at Alibaba Cloud (also known as Tongyi Qianwen lab).
- Imagen 3v6Imagen 3 is a text-to-image generation model developed by Google DeepMind, announced at Google I/O on May 14, 2024 and progressively rolled out to users through mid-2024 and into 2025.
- Kling 2.1v5Kling 2.1 is a video generation model developed by Kuaishou Technology, the Beijing-based internet and short-video company behind China's second-largest short-video platform.
- GLM-4.5v6GLM-4.5 is an open-weights large language model released by Zhipu AI (operating internationally as Z.ai) on July 28, 2025, built on a 355-billion-parameter Mixture of Experts architecture that activates 32…
- FLUX.1v5FLUX.1 is a family of text-to-image generation models developed by Black Forest Labs, released on August 1, 2024.
- Hume AIv4Hume AI is a New York-based artificial intelligence research company and API platform, founded in March 2021 by Alan Cowen, that builds "emotionally intelligent" voice AI: models trained to measure human…
- Kimi K2v5Kimi K2 is an open-weights Mixture of Experts language model from Moonshot AI, a Beijing startup, released on July 11, 2025 with 1.04 trillion total parameters and 32.6 billion activated per token
- Vapiv7Vapi is a voice AI orchestration platform that lets software developers build, deploy, and scale AI phone agents through a programmable API.
- Sesame (AI company)v5Sesame (formally Sesame AI Labs) is a San Francisco-based artificial intelligence company founded in June 2023, best known for developing the Conversational Speech Model (CSM) and the Maya and Miles voice…
- Letta (MemGPT)v5Letta (formerly MemGPT) is an open-source platform for building stateful AI agents with persistent memory, developed by the team behind the MemGPT research project at UC Berkeley's Sky Computing Lab.
- Cartesiav5Cartesia is a San Francisco-based AI company focused on real-time voice synthesis, speech recognition, and state space model (SSM) research.
- Retell AIv4Retell AI is a voice AI agent platform that enables businesses to build, deploy, and manage AI-powered phone agents for inbound and outbound call automation.
- Agent2Agent Protocolv5The Agent2Agent (A2A) Protocol is an open standard for communication and interoperability between independent AI agents.
- Mem0v6Mem0 (pronounced "mem-zero") is an open-source memory layer for AI agents and large language model applications that gives stateless LLMs persistent, evolving memory across sessions.
- LanceDBv4LanceDB is an open-source, developer-friendly vector database and multimodal lakehouse built on the Lance columnar storage format, designed to store vector embeddings, images, video, audio, and structured…
- Browser Usev4Browser Use is an open-source Python library that lets large language models control a web browser, turning any LLM into an agent that can browse the web, click buttons, fill forms, and complete multi-step…
- Roo Codev3Roo Code is an open-source AI coding agent that runs as a Visual Studio Code extension.
- Kiro (AI IDE)v4Kiro is an agentic integrated development environment (IDE) built by Amazon Web Services and released in public preview on July 14, 2025 .
- Lindyv5Lindy is a no-code AI agent platform that lets businesses and individuals build autonomous agents to automate workflows across sales, customer support, operations, and scheduling.
- Augment Codev4Augment Code is an enterprise AI coding platform, founded in 2022 by Igor Ostrovsky and Guy Gur-Ari and led by CEO Scott Dietzen, that builds AI agents purpose-built for large, complex codebases.
- GraphRAGv3GraphRAG is a graph-based approach to retrieval-augmented generation developed by Microsoft Research, first described publicly on February 13, 2024 and formalized in the paper "From Local to Global: A Graph…
- Gumloopv3Gumloop is a no-code AI workflow automation platform that lets teams build, deploy, and manage AI agents and automated workflows through a visual drag-and-drop interface.
- n8nv4n8n (pronounced "n-eight-n") is a source-available workflow automation platform built for technical teams, developed by the Berlin-based company n8n GmbH.
- Robotv5A robot is a programmable machine that senses its environment, computes a decision, and acts on the physical world through a closed loop of perception, planning, and actuation.
- Gemini 2.5 Prov6Gemini 2.5 Pro is the flagship reasoning large language model of Google's Gemini 2.5 family, developed by Google DeepMind and first released as an experimental preview on March 25, 2025
- GPT-5.5v6GPT-5.5 is a large language model developed by OpenAI and released on April 23, 2026.
- Claude Opus 4.5v7Claude Opus 4.5 is a large language model developed by Anthropic and released on November 24, 2025.
- RT-2v4RT-2 (Robotic Transformer 2) is a vision-language-action model developed by Google DeepMind that enables robots to execute novel tasks by transferring knowledge from internet-scale vision-language pretraining…
- Chelsea Finnv4Chelsea Finn (born October 8, 1992) is an American computer scientist, an assistant professor of computer science and electrical engineering at Stanford University, and a co-founder of the robotics company…
- Sony Group Corporationv3Sony Group Corporation (Japanese: ソニーグループ株式会社, Sonī Gurūpu Kabushiki gaisha) is a Japanese multinational conglomerate that is the world's leading supplier of CMOS image sensors (with a revenue share near 50…
- Tencentv8Tencent Holdings Limited (Chinese: 腾讯控股有限公司; pinyin: Téngxùn) is a Chinese multinational technology and entertainment conglomerate based in Shenzhen, Guangdong, and is one of the world's largest internet…
- Stuart Russellv4Stuart Jonathan Russell (born 1962) is a British computer scientist, professor of computer science at the University of California, Berkeley, and one of the most influential figures in modern artificial…
- Harvard Universityv4Harvard University is a private Ivy League research university in Cambridge, Massachusetts, founded in 1636, whose principal stake in artificial intelligence is the Kempner Institute for the Study of Natural…
- Paul Allenv3Paul Gardner Allen (January 21, 1953 to October 15, 2018) was an American business magnate, computer programmer, investor, and philanthropist who co-founded Microsoft with his childhood friend Bill Gates in…
- Pieter Abbeelv5Pieter Abbeel (born 1977) is a Belgian-American computer scientist and a professor of electrical engineering and computer sciences at the University of California, Berkeley, where he directs the Berkeley Robot…
- Siemensv4Siemens AG is a German multinational technology conglomerate, headquartered in Munich, that is the largest industrial manufacturing company in Europe by revenue and one of the world's leading providers of…
- Ion Stoicav3Ion Stoica is a Romanian-American computer scientist, professor of electrical engineering and computer sciences at the University of California, Berkeley, and a serial entrepreneur whose academic and…
- π0v4π0 (pronounced "pi-zero") is a vision-language-action model for general-purpose robot control developed by Physical Intelligence, a San Francisco-based robotics startup, and introduced on October 31, 2024.
- Pattern Recognitionv3Pattern recognition is the automatic discovery of regularities in data through the use of computer algorithms
- Twin Delayed DDPGv4Twin Delayed Deep Deterministic Policy Gradient (TD3) is an off-policy actor-critic reinforcement learning algorithm for continuous action spaces, introduced by Scott Fujimoto, Herke van Hoof, and David Meger…
- Advanced Driver-Assistance Systemsv4Advanced Driver-Assistance Systems (ADAS) are electronic technologies that help drivers operate, steer, brake, and park a vehicle by warning of hazards or taking momentary or sustained control of the driving…
- SwiGLUv6SwiGLU (Swish-Gated Linear Unit) is the activation function used inside the feed-forward sublayer of most modern transformer large language models, including LLaMA, PaLM, Mistral, Qwen, and DeepSeek.
- Reactv4React (sometimes called React.js or ReactJS) is a free, open source JavaScript library for building user interfaces, created by Jordan Walke at Facebook and first released publicly on May 29, 2013.
- Sergey Levinev3Sergey Levine is an American computer scientist, associate professor of electrical engineering and computer sciences at the University of California, Berkeley, and a co-founder of Physical Intelligence
- University of California, Berkeleyv5UC Berkeley is the public research university most responsible for the open-source infrastructure that runs modern AI: its systems labs created Apache Spark, Ray, and vLLM, and its Berkeley Artificial…
- Redwood Researchv6Redwood Research is a nonprofit AI safety organization founded in 2021 and headquartered in Berkeley, California, best known for pioneering the "AI control" research paradigm and for its landmark December 2024…
- Bill Gatesv4William Henry Gates III (born October 28, 1955), known as Bill Gates, is an American businessman, software developer, philanthropist, and prominent public commentator on artificial intelligence.
- Gopher (language model)v3Gopher is a 280-billion-parameter autoregressive transformer language model developed by DeepMind and described in a trio of companion papers released on December 8, 2021.
- Horovodv3Horovod is an open-source distributed training framework for deep learning that lets a single-GPU training script scale across many GPUs and many machines by adding only a few lines of code.
- Cari Tunav3Cari Tuna (born October 4, 1985) is an American philanthropist and former Wall Street Journal reporter who, with her husband Dustin Moskovitz, co-founder of Facebook and Asana, founded the philanthropic…
- Next.jsv3Next.js is an open-source React framework developed and maintained by Vercel that has become the default way to ship production AI products, especially streaming chat interfaces and other large language model…
- Holden Karnofskyv3Holden G. Karnofsky is an American philanthropist, nonprofit executive, and writer on artificial intelligence who co-founded the charity evaluator GiveWell (2007) and the grantmaking organization Open…
- International Conference on Computer Visionv4The International Conference on Computer Vision (ICCV) is one of the three top-tier academic conferences in computer vision, held every two years in odd-numbered years since 1999 and first staged in London in…
- Soft Actor-Criticv3Soft Actor-Critic (SAC) is an off-policy, maximum-entropy deep reinforcement learning algorithm that trains a stochastic actor-critic to maximize expected reward plus the entropy of its own policy, so the…
- Apple Inc.v4Apple Inc. is an American multinational technology company headquartered in Cupertino, California.
- Python (programming language)v5Python is the dominant programming language for artificial intelligence and machine learning: roughly 59 percent of research implementations tracked by Papers with Code in September 2024 used PyTorch, a Python…
- OLMov4OLMo (Open Language Model) is a family of fully open large language models built by the Allen Institute for AI (Ai2) and first released on February 1, 2024.
- Dan Hendrycksv4Dan Hendrycks (born 1994 or 1995) is an American machine learning researcher who serves as executive director of the Center for AI Safety, the San Francisco nonprofit he co-founded in 2022, and is the lead…
- Paul Christianov5Paul Christiano is an American AI safety researcher who is one of the principal architects of Reinforcement Learning from Human Feedback (RLHF), the technique used to fine-tune ChatGPT, Claude, and most modern…
- Normal distributionv4The normal distribution, also called the Gaussian distribution, is a continuous probability distribution defined by two parameters, a mean $$\mu$$ and a variance $$\sigma^2$$
- Gary Marcusv3Gary F. Marcus (born 1970 in Baltimore, Maryland) is an American cognitive scientist, author, and entrepreneur who is the best known public critic of contemporary deep learning and large language models, and a…
- OMRON Corporationv3OMRON Corporation is a Japanese electronics, industrial automation, and healthcare technology company headquartered in Shimogyo-ku, Kyoto, and is one of the world's larger suppliers of factory automation…
- HotpotQAv3HotpotQA is a large-scale, multi-hop question answering dataset of about 112,779 crowd-authored question-and-answer pairs over English Wikipedia, whose answers cannot be found in any single paragraph and…
- EXAONEv3EXAONE (an acronym for EXpert AI for EveryONE) is the family of large language models and foundation models developed by LG AI Research
- EtherCATv3EtherCAT (Ethernet for Control Automation Technology) is a real-time industrial Ethernet fieldbus protocol, developed by Beckhoff Automation and introduced in April 2003, that delivers microsecond-class…
- Robert Bosch GmbHv3Robert Bosch GmbH, known as Bosch, is a German engineering and technology company that is the world's largest automotive supplier and one of Europe's largest corporate investors in artificial intelligence.
- Apache MXNetv3Apache MXNet (pronounced "mix-net") was an open-source deep learning framework that combined imperative and symbolic execution in one runtime, created around 2015 by the DMLC (Distributed Machine Learning…
- COMPAS (recidivism risk assessment)v3COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) is a proprietary actuarial risk-assessment instrument used by United States courts and corrections agencies to estimate the…
- European Conference on Computer Visionv3The European Conference on Computer Vision (ECCV) is the biennial top-tier academic conference on computer vision, held in even-numbered years at locations across Europe, and is one of the three most…
- SmoothGradv3SmoothGrad is a saliency map technique that reduces visual noise in gradient-based explanations of neural network predictions by averaging gradients over many noisy copies of the input.
- RobotEra L7v2The RobotEra L7 (Chinese: 星动 L7, Xingdong L7) is a full-size bipedal humanoid robot developed by Beijing-based RobotEra (星动纪元), a Tsinghua University spin-off founded in August 2023.
- RefinedWebv4RefinedWeb is a large-scale English pretraining dataset for large language models, built from filtered and deduplicated Common Crawl web data alone and released in June 2023 by the Technology Innovation…
- Winograd Schema Challengev3The Winograd Schema Challenge (WSC) is a commonsense reasoning test in which a system must resolve an ambiguous pronoun in a short sentence where the correct answer flips when one or two words change
- Dolmav6Dolma is an open three-trillion-token English pretraining corpus released by the Allen Institute for AI (AI2) to power its fully open OLMo language models and to let researchers study how training data shapes…
