AI Glossary Defines Models, Labs, and Clients
0xdf, a security researcher and CTF blogger, published an AI glossary (a cheatsheet) that defines core terminology for practitioners tracking the fast-moving AI ecosystem. The reference distinguishes frontier AI labs (which keep weights private and expose models via APIs or official clients) from open-weight labs (which release downloadable weights), and separates models from the clients and harnesses that wrap them. It defines an LLM as the foundational model class, a model as a versioned trained artifact made of numerical weights, and introduces tokens as the unit models use to process text. It lists representative labs such as Anthropic, OpenAI, Google DeepMind, and xAI, and open-weight players such as DeepSeek, Meta, and Mistral, grouped by capability and cost rather than version numbers alone. It is a practitioner reference rather than news.
What it is
0xdf, a security researcher known for CTF and home-lab writeups, published an AI glossary, structured as a cheatsheet, that lays out current AI vocabulary for practitioners. It distinguishes frontier AI labs, which keep their model weights private and require access through an API or an official client, from open-weight labs, which release weights that anyone can download, host, or fine-tune. The glossary stresses that "open weight" is not the same as "open source": only the trained weights are downloadable, while training data, code, and exact process usually remain private. It also separates the entities that produce models (labs) from the systems that wrap them (clients and harnesses).
What it defines
The reference describes an LLM as the foundational model class in current practice and a model as a named, versioned trained artifact composed of numerical weights, and it begins explaining tokens as the unit of text models process. It lists representative frontier labs such as Anthropic, OpenAI, Google DeepMind, and xAI, and open-weight players such as DeepSeek, Meta, and Mistral, and groups model families by capability and cost trade-offs rather than by version number alone, mirroring how teams actually choose between hosted APIs and self-hosted open-weight models.
Why it matters
Rapid naming proliferation has increased friction for practitioners mapping capabilities to engineering choices. A concise taxonomy of labs, models, clients, and initiatives reduces ambiguity when evaluating benchmark claims, licensing constraints, and deployment paths, and improves documentation quality for teams maintaining model inventories. It is a useful onboarding and reference artifact rather than new research or a product release.
Key Points
- 10xdf's AI glossary defines labs, models, clients, and tokens, separating weight-private frontier labs from open-weight labs that release downloadable models.
- 2Naming proliferation and overlapping terms create real friction; a concise shared vocabulary reduces ambiguity in benchmarking, licensing, and deployment choices.
- 3For teams maintaining model inventories or procurement catalogs, consistent definitions improve comparability and documentation, though it adds no new research.
Scoring Rationale
A clear, practical reference cheatsheet that helps practitioners disambiguate AI terminology (labs vs models vs clients, frontier vs open-weight), but it presents no new research, benchmark, or product. As a single-author explainer it is genuinely useful for onboarding yet low in news weight, placing it in the minor-but-relevant band.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
