Learn

• For anyone starting from scratchNew to AIWhat a language model is, how it reads and writes, what happens to what you type, and how to look up a model before you use it.
  1. 1How large language models work
  2. 2Tokens and context windows
  3. 3Multimodal models: text, images, audio, and video
  4. 4What happens to what you type into an AI service
  5. and 2 more
  6 steps · about 28 min of readingStart
• For people who want to download or self-host a modelChoosing and running a modelRead the license, check the model card, decide between hosted and local, and work out whether a model fits your hardware.
  1. 1What open weight and open source actually mean
  2. 2How to read a model card
  3. 3Choosing hosted access or local inference
  4. 4Understanding inference hardware
  5. and 3 more
  7 steps · about 32 min of readingStart
• For developers and technical teamsBuilding with AIHow models call tools, answer from your documents, run as agents, and get adapted through fine-tuning.
  1. 1How AI models use tools
  2. 2Retrieval-augmented generation (RAG)
  3. 3Agents and robotics without the hype
  4. 4Pretraining, fine-tuning, and post-training
  5. and 2 more
  6 steps · about 23 min of readingStart
• For journalists, researchers, students, and policy readersChecking claims about AIRead benchmark results, safety documents, content labels, data disclosures, infrastructure announcements, and policy texts with care.
  1. 1How to check an AI claim
  2. 2How to read an AI evaluation
  3. 3How AI developers test models for safety
  4. 4Labels, watermarks, and content credentials
  5. and 4 more
  8 steps · about 48 min of readingStart
• For readers who want the big pictureUnderstanding the industryWhere today's AI came from, how chips, compute, labs, and software fit together, and the work people do in the field.
  1. 1A short history of AI in the United States
  2. 2How the American AI ecosystem fits together
  3. 3Foundation models and the labs that build them
  4. 4Chips, cloud, and compute
  5. and 2 more
  6 steps · about 16 min of readingStart

Explainers by topic

Each answers one question in a few minutes, with a worked example and a list of the sources it was checked against.

Terms to know

Plain definitions, each with how the catalog uses the term and a real example. The glossary has all 69.

• Weights
• Token
• Inference
• Context window
• Open weight
• Model card
• Fine-tuning
• Quantization
• Hallucination
• Embedding
• Retrieval-augmented generation (RAG)
• Agent
• Function calling (tool calling)
• Model Context Protocol (MCP)
• Benchmark
• Red teaming

Topic hubs

Sourced introductions to parts of the landscape, each with a reading path into the catalog and its primary documents.

• AI agents and agent toolingWhat separates an agent from a model, and the open protocols, SDKs, and coding agents used to build agents, including the Model Context Protocol, A2A, and AGENTS.md.
• AI for scienceBiology, chemistry, and medical models in the catalog, the science-focused organizations that publish them, and the licenses, access gates, and safety measures their documents describe.
• AI in enterprise softwareEnterprise software and data-platform companies in the catalog, the AI products they document, the outside models those products can call, and the open releases some of them publish under separate terms.
• AI safety, security, and trustHow developers, standards bodies, and independent groups test AI systems for harm and misuse: safety frameworks and model cards, testing tools, safety classifiers, content provenance and watermarking, and prompt injection.
• Benchmarks and evaluationEvaluation harnesses, benchmarks, and the organizations that run them, and what a published result can and cannot tell you about who tested what, on which version, and under which conditions.
• Building software with AIThe open-source layers developers use to build with AI models: model libraries, inference servers, application SDKs and frameworks, coding agents, and experiment tracking, and what to check about where each one sends data.
• Chips, cloud, and computeThe hardware and services behind AI: chip designers, accelerators, memory, networking, and AI cloud providers, and why a headquarters is not a data-center location.
• Foundation models and the labs that build themHow model developers in the catalog make their models available, hosted, as downloadable weights, or both, why a model family is not a release, and how to read access and license terms.
• Open-source foundations and stewardsHow foundations such as the Linux Foundation, the PyTorch Foundation, LF AI & Data, and the Agentic AI Foundation host AI projects, and why a nonprofit's tax status is separate from the openness of what it hosts.
• Robotics and embodied AIRobot makers, robot foundation models, simulators, and driving data in the catalog, with the licenses they carry and the testing and safety limits their own documents state.
• Running AI on your own hardwareHow running AI models on your own computer or device works: local runtimes, on-device frameworks, quantization, and what to check about an open-weight model before you run it.
• Speech, vision, and multimodal AIModels and tools that work with images, audio, speech, and video: image-text encoders, vision-language models, speech recognition, segmentation and vision backbones, and image and video generation, with the terms that come with them.
• Training data and datasetsOpen pretraining datasets in the catalog and the organizations that steward them, how many of them build on Common Crawl, and why a dataset's license and documentation are not permission to reuse what it contains.

Look things up

• Frequently asked questionsShort, sourced answers with links to go deeper
• GlossaryTerms used across the catalog
• License guideWhat the licenses on models, software, and datasets say
• U.S. AI policy trackerFederal laws, orders, memoranda, and standards on AI
• Model size calculatorHow much storage a model's weights take at each precision
• PlacesOrganizations by documented headquarters state
• TimelineOpen releases and news events by documented date
• Source libraryEvery cited source and the records that cite it
• Data and reuseThe public data files and reuse terms
• MethodologyEligibility, openness, sources, and dates